11.

A.
. .1
1 . .11.?
«.31...» .
. n _

, . .1}...
. 1 if...
. . . «Rant E
3.4. , at...» r
. L‘I... . . 1. I. .
., . , .25. . $4: .1... .
.fifiun‘ , .1. . 1 . 1. . .
t..rlv‘..r . . t . ,
i... 511:
1.1.. 31:1,...411. _ 1
1 . ”an...
.I .

:3 a.

1 b. h"?

I -
‘ianu .

‘1}:

v.7?

(u’rnm
C

I: . 74.x
.1 31} 1.. . «SW
‘ ix. 2 . . . . 1
:3... fiqgntvfiuia _ .5
1 . .. S... 1
‘3’}: .
.12....

1:3.
I:

PI}:
9.3:: .
Z.
. .1.
3i» . a. h
; = s.
1... ‘ 11...}:

in»...

“1%.?

1i! 3......

V£U“
. n ,

, .53.
:HPJHPRA. , .Qi a;
1. 0-... .Q a x ..

eaaamfiqfimw

. up, Inuwmn<iné
gnaw
g .

1
31321.... . . .2.
.3 1.31:: 22:6 33:: 32:09?

1.. x. u E... .33.:6‘9!‘ I.)
. 1 .1 .7... I 1. . 3111;11:322.

drum

(1. .
. r. (a... l.
kayak? l
1 ’lceh )0), i-I 5“ it ‘Qi’iip
. it. 3.5 I ’31.»; rig E
fl 1:1! .2: 1:! 4i"! .
1.. t .1: In...
C‘
a

 

.69...

. . ‘1‘}...13 191......
:15}... .r .7). 3... 1:2
~‘v.# .0. .491 tn...’ Q’- 734’
. .1. - V

1..
11?... xvii...

 

 

 

. .w

.16..

wags

 

/‘"\.'4'

" ‘ unnnnv
Mun-.3..- male

Unlverslty

 

 

 

This is to certify that the

dissertation entitled

Selective Exposure, Uses and Gratifications
of a Cyber Election Campaign:
Presidential Election 2000

presented by

Yasuhiro Inoue

has been accepted towards fulfillment
of the requirements for

Ph.D. Mass Media

degree in

 

 

Anagkm A W57

Major professor

319/200!

Date

 

MSU i: an Affirmatiw Action/Equal Opportunity Institution 0-12771

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/ClRC/DateDue.pBS—p. 15

Selective Exposure, Uses and Gratifications of a Cyber Election
Campaign: Presidential Election 2000

By

Yasuhiro Inoue

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Mass Media Ph.D. Program
College of Communication Arts and Sciences

2001

Abstract

Selective Exposure, Uses and Gratifications of a Cyber Election
Campaign: Presidential Election 2000

By

Yasuhiro Inoue

This dissertation examines factors that determine voters’ use of candidates’
online campaigning and the Web’s possible effects on voting behavior in the Presidential
Election 2000 by applying selective exposure and uses and gratifications theories. The
specific purpose of this dissertation is five-fold; this study (1) tests whether voters are
more likely to log on to a preferred candidate’s Web site, (2) investigates whether
selective exposure to a preferred candidate’s Web site affects the endorsement for the
candidate, (3) judges whether Web site use enhances voting turnout, (4) explores why
voters log on to online campaign sites; and (5) examines if there are differences in voters’
gratifications between online campaigning and other media.

These objectives were accomplished by conducting a two-wave panel survey of
college students. Survey questionnaires were administered to 461 college students at
Time 1 (September 2000) and 366 at Time 2 (November 2000). Time 1 survey
consisted of voting decisions for a candidate, uses and gratifications and demographic
profiles. Time 2 survey asked respondents about the frequency of logging on to
candidate Web sites, news Web sites and whether and whom they voted for on election
day. The survey explored the relationship between voters’ preference and exposure to
online campaigning over time, as well as the possible effects of online campaigning on

voting behavior. It also mapped what gratifications voters sought from campaign Web

sites and how they made use of such sites.

The dissertation documented five major findings: (1) a potential for Web sites to
enhance the probability of voting turnout, (2) respondents’ tendency of selective
exposure toward preferred candidate Web sites, (3) third party supporters’ dependency on
candidate Web sites, (4) the media-specific nature of political uses and gratifications and
(S) respondents’ tendency to be gratified more by the old media than by the online media.

The first major finding provided empirical evidence contrary to what had been
found in the previous studies, i.e., the online media were predictors for voting turnout
while the old media were not. This finding allowed me to construct the cyber
motivation hypothesis, which posits that (1) selective exposure to consonant messages on
a Web site enhances confidence, or self efficacy, (2) the Web site offers a sense of
community which enhances self-efficacy, and (3) the obtained and enhanced self-efficacy
eventually affects voting behavior, i.e., motivates people to vote.

This dissertation provided new empirical evidence about online election
campaigning and cast light on voters’ uses and gratifications on candidate Web sites.
However, due to the lack of representation and ever changing role of the Internet in
politics, it is necessary for researchers to constantly update the roles of the Internet in

presidential election campaigns.

Cow/fight by
Yasuhiro Inoue
2001

Dedicated to my father and mother,

Yutaka and Kayoko Inoue

ACKNOWLEDGEMENTS

I appreciate my family members, Maria, Anna, Shinnosuke and especially my
wife Shoko for their patience and love throughout my almost six years in the United
States. With Shoko’s optimism, encouragement and sacrifice, I was able to start study-
ing at Michigan State University and complete the Ph.D. My parents and grandparents
always understood whatever I have done throughout my life. My family is great.

I could not have completed my degree without the help of numerous people. I
would like to express my gratitude and appreciation to my principal advisor and mentor,
Dr. Lucinda Davenport, for her guidance and support throughout not only this disserta-
tion writing but also three years in the Mass Media Ph.D. program. I would also like to
thank the members of my committee, Dr. Howard Bossen, Dr. Jim Dearing, and Dr. Den-
nis Patterson. Dr. Bossen’s media technology class gave me a new perspective with
which I worked on this dissertation. Dr. Dearing guided me throughout my academic
life here at MSU. Dr. Patterson was of great help for instructing me in political science
field which had not been my specialty.

Amy Pavlov and Penny Owen, my dissertation writing group consultant and
partner, were not only my writing teachers but also my collaborators. They deserve the
credit that I was able to write this dissertation in an organized manner. I owe greatly to
my professors and friends who were kind enough to help collect data for me, Dr. Bossen,
Dr. Patterson, Dr. Hal Bush, Dr. Mary Bresnahan, Dr. Tom Kiyomiya, Jay Newell, and
Marina Choi. My friends Dr. Yasuo Miyazaki and Sachiyo Morinaga helped me ana-
lyze the data. My neighbor Alison Steiber “deciphered” respondents’ illegible hand-

writing for me. Thank you all!

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. ix
LIST OF CHARTS AND FIGURES .................................................................................. x
CHAPTER I
INTRODUCTION .............................................................................................................. 1
Justification ........................................................................................................... 1
Cyber Election Campaigning ................................................................................ 4
Candidate Web Sites and Descriptions .................................................. 7
CHAPTER II
THEORETICAL FRAMEWORKS .................................................................................. 13
Selective Exposure .............................................................................................. 13
Partisan/Reinforcement Perspectives .................................................. 15
Schemata Perspective .......................................................................... l9
Cognitive Consistency (Dissonance) Perspective ............................... 20
Voter Behavior ........................... , ......................................................................... 23
Third Party Supporters ........................................................................................ 28
Uses and Gratifications ....................................................................................... 30
CHAPTER III
METHOD ......................................................................................................................... 37
Research Desrgn37
Questionnaire and Variables ................................................................................ 39
Voter Preference/Partisanship Variables .............................................. 40
Media Exposure Variables ................................................................... 4O
Gratifications Variables ....................................................................... 41
Control Variables ................................................................................. 42
Analysis Method ................................................................................................. 43
CHAPTER IV
RESULTS .......................................................................................................................... 45
Participants .......................................................................................................... 45
Information Media Use ....................................................................... 45
Voting .................................................................................................. 46
Selective Exposure .............................................................................................. 46
Voting Behavior .................................................................................................. 62
Third Party Supporters ........................................................................................ 70
Uses and Gratifications ....................................................................................... 74
Summary of Results ............................................................................................ 88

vii

CHAPTER V

DISCUSSION ................................................................................................................... 91

Possible Impact on Voting Behavior ................................................................... 91

Cyber Motivation Hypothesis ............................................................. 93

Selective Exposure to Candidate Web Sites ....................................................... 95

Third Party Supporters’ Web Use ....................................................................... 97

Media Specific Uses and Gratifications .............................................................. 98
CHAPTER VI

CONCLUSIONS ............................................................................................................. 102

Contributions and Appropriateness ................................................................... 102

Implications on Presidential Election Campaigning ......................................... 105

Limitation and Recommendations .................................................................... 107

APPENDIX A ................................................................................................................. 110

APPENDIX B ................................................................................................................. 114

APPENDIX C ................................................................................................................. 115

REFERENCES ............................................................................................................... 122

viii

Table 1
Table 2

Table 3

Table 4

Table 5

Table 6
Table 7
Table 8

Table 9

Table 10

Table 11

Table 12
Table 13

Table 14

LIST OF TABLES

Media Richness of the Presidential Candidate Web Sites ................................... 9
Mean Frequencies of Visiting Each Candidate Web Site by Bush, Gore and
Nader Supporters ............................................................................................... 50
Percentages of Visiting Each Candidate Site by Bush, Gore and Nader
Supporters .......................................................................................................... 52
Logging on Frequencies of Each Candidate’s Supporters by Weak, Middle and
Strong Support Degrees .................................................................................... 57
Logging on Probabilities for Weak, Middle and Strong Supporters by Each .
Candidate Supporters ........................................................................................ 59
Voting Behavior by Online Information Use .................................................... 63
Logit Regression Analysis about FactOrs Determining Voting Behavior ......... 66
Probability of Using Candidate Web Sites by Major Party Candidate
SupportersNoters and Third Party Supporters/Voters ....................................... 72
Probability of Using News Web Sites by Major Party Candidate
Supporters/Voters and Third Party Supporters/Voters ....................................... 73
Means Values and Standard Deviations of 11 Gratifications Items

by Each Media ................................................................................................... 76
Factor Analysis of Gratifications items for Presidential Election

Information Sources .......................................................................................... 78
Correlation Matrix between Gratifications Factors and Media Use ................. 84

Regression Analysis Results Indicating Impact of Gratifications on
Candidate and News Web Site Use 86

Index Means and Major Party Candidate SupportersNoters v. Third Party
Candidate Supporters/Voters ............................................................................. 87

ix

LIST OF CHARTS AND FIGURES

Chart 1 Bush Supporters’ Candidate Web Site Use ....................................................... 47
Chart 2 Gore Supporters’ Candidate Web Site Use ....................................................... 48
Chart 3 Nader Supporters’ Candidate Web Site Use ...................................................... 49
Chart 4 Bush Supporters’ Logging on Frequency by Support Degree .......................... 54
Chart 5 Gore Supporters’ Logging on Frequency by Support Degree ........................... 55
Chart 6 Nader Supporters’ Logging on Frequency by Support Degree ......................... 56
Chart 7 Odds for Voting as Evidenced by Online Information Use ............................... 68
Chart 8 Probabilities for Voting as Evidenced by Online Information Use ................... 69

Figure 1 Selective Exposure and Web Sites’ Booster Effect Model,
Cyber Motivation Hypothesis ........................................................................... 94

CHAPTER I

INTRODUCTION

Justification

In the presidential election of 2000, every candidate jumped onto the online
campaign bandwagon. Online campaigning made its debut in the 1996 presidential race,
but was used only as a novelty. Some argue that online campaigning in 2000 was still a
“beta version” and did not carry substantial implications (Dickerson, 2000). While
television is said to have contributed to John F. Kennedy’s victory over Nixon in the
1960 presidential election, the Internet has not played a crucial role in presidential elec-
tions. There is no “J .F.K. of the Internet” yet (Reed, 1999).

Online campaigning, however, may have the potential to change the way candi-
dates campaign. The Internet allows candidates to communicate with voters, strengthen
support, raise funds, and recruit volunteers in ways that other media do not. ~ Some ar-
gue that the Internet will change the whole process of politics including election cam-
paigns, policymaking, and perhaps even the balance of political power in the near future
(Hill &'Hughes, 1998; Morris, 1999).. ' The composite of media channels for modern
election campaigns may soon be transformed due to the utilization of the Internet.

While the utilization of the Internet by politicians has been documented and ap-
plied to actual campaign strategies, our understanding of how and why voters use online
campaign information is still vague. Because the Internet provides three unique fea-
tures that will empower voters, it is worth investigating online campaigning from the
standpoint of the voters. One of the unique features is that information on the Web is of

the “pull” variety, which means that the selection of online information is controlled by

Web users. Users can immediately “pull” information from online media at their con-
venience, while information in newspapers and on television is “pushed” to the readers
and viewers. The second feature, customization, allows users to tailor and/or screen out
online information for their purpose, whereas information in newspapers and on televi-
sion is fixed. The third feature is the Intemet’s interactivity, which makes it possible for
voters to participate actively in political life by signing up to volunteer, to contribute
money online to a candidate and to interact with those who share common interests.
These three feamres of the Internet should have tremendous implications for
election campaigns and the future of democracy in a way similar to the invention of the
printing machine in the 15th century made possible. the mass-production of books and the
spread of literacy, which eventually led to profound social and political changes. For
example, voters and small political. groups alike will be able to counterbalance the flows
of political information controlled by politicians and to accomplish true or ideal forms of
democracy by increasing their participation in the political process via the Internet.
Specifically, before the inception of the Internet, voters were provided only with infor-
mation filtered by the mainstream media so“ that they could not obtain information about
marginalized political activists. Now, at least technically the US. president and a third
party candidate, and even a ordinary person, have the same means of communication to
the same potentially large audience (Hill ‘& Hughes, 1998). Everybody can have a
voice on the Internet and third party supporters can obtain information online that was
nearly unavailable in the mainstream media. Thus, it is equally as important to under-
stand why people choose to visit a- specific Web site, for what purpose, and under what
predispositions as it is to understand what effect the Internet information has.

In the light of these implications, this study will examine factors that determine

voters’ use of online campaigning and the Web’s possible effects on voting behavior in
the Presidential Election 2000 by applying selective exposure and uses and gratifications
theories. Selective exposure refers to a tendency toward supportive messages in order
to increase confidence about thought and behavior. Uses and gratifications framework
explains how people use media to gratify their needs and understand their motives for
media behavior. The presidential election 2000 offered a special opportunity to inves-
tigate voters’ use and possible effects of the Internet.

The specific purpose of this paper is five-fold; this study (1) tests whether voters
are more likely to log on to a preferred candidate’s site, (2) investigates whether selective
exposure to a preferred candidate’s site affects the endorsement for the candidate, (3)
judges whether Web site use enhances voting turnout, (4) explores why voters log on to
online campaign sites; and (5) examines if there are differences in voters’ gratifications
between online campaigning and other media.

This study accomplishes these objectives by conducting a two-wave panel sur-
vey of college students. Survey questionnaires were administered to 461 college stu-
dents at Time 1 (September 2000) and 366 at Time 2*(November 2000). The panel sur-
vey explores the relationship between voters’ preference and exposure to online carn-
paigning over time, as well as the possible effects of online campaigning on voting be-
havior. It also maps what gratifications voters seek from campaign Web sites and how
they make use of such sites. This study expects to find distinct roles of Web sites in the
election campaigning because of the unique features of the Internet. Understanding the
selection and needs of voters will provide more insight into the study of online election
campaigning and cast light on new and growing voters’ digitalroles in democracy in the

. 2lst century.

Cyber Election Campaigning

The Internet is dramatically changing the communications environment. An
ever increasing number of people go online and communicate by email. The diffusion
rate of Internet access among American people is growing rapidly. According to Pew
Internet & American Life Project (2001), while 47 % of all adult Americans had Internet
access in May-June 2000, a half-year later, the percentage increased to 56%. The num-
ber of email users also increased from 44% in May-June 2000 to 49% in Novem-
ber-December 2000. Beside e-mail, there are a variety of ways people are utilizing the
Internet. “Hunt for hobby information” was the most popular feature (79%) among
Internet users followed by “browse for fun” (68%). “Obtain online news” was the third
purpose (63%). “Online shopping” was ranked in fifth place (52%) following “obtain
medical information” (57%).

The Internet is changing not only the way people communicate, obtain informa-
tion and do business today (Korgaonkar & Wolin,‘ 1999), but also the way candidates
campaign, and the way voters participate in politics (Morris, 1999). It can even change
the way voters cast ballots. Computer companies and universities are now developing
Internet voting systems, “e-voting,” which incorporate voter registration, voter identifi-
cation and ballot casting and counting (Schwartz, 2001). Some of the ballots in the
2000 presidential election were cast over the Internet before the election by 200 service
people stationed overseas in a pilot program developed by the US. Department of
Defense (Thomas, 2000).

It is said that 1996 marked the year that candidates began campaigning online
(Whillock, 1997). For example, all major presidential candidates and two thirds of ma-

jor party senatorial candidates established campaign Web sites, while only one fifth of

major party congressional candidates did so (D’Alessio, 1997). However, few candi-
dates used the Internet strategically for campaigning (Reed, 1999). Candidates’ Web
sites were so primitive that few candidates offered online contribution and e-mail net-
works at that time (Zeller, 2000). In addition, Internet users composed a relatively mi-
nor segment of the US. population at that time. According to a survey result, only 23%
of adult Americans experienced “going online” in 1996 (Pew Research Center, 2001).
Thus, it was uncertain what online campaigning could do and how effective it would be
in the 1996 race (Whillock, 1997).

Yet, the year 1998 witnessed dramatic growth in the number of campaign sites
and voters’ attention to campaign sites, probably due to the significant diffusion of the
Internet to US. citizens. Scholars and political consultants then started to seriously
examine the potential impact of online campaigning on voters. For example, Reavy

(1999) explains the intended mechanism of online campaigning as follows:

The lntemet gives candidates the ability to disseminate information about them-
selves. Citizens who possess knowledge about their politicians play a more ac-
tive role in determining which of those politicians shall govern and whiCh shall
be governed. It follows that those who gain information from the Internet. ..
are more actively engaged in the political process than those who do not (p.
249) '

To paraphrase Reavy, online campaigning can turn interested voters into in-
formed voters and informed voters into volunteers. The lntemet allows voters to par-
ticipate in politics and have politicians hear the voters’ voices. Candidates began rec-
ognizing these roles and potentials of the lntemet in the late 19903.

One of the great advantages of online campaigning is that it can mobilize and

organize supporters regardless of geographical distance (Dickerson, 2000). In the 1998
election, current Minnesota governor, Jesse Ventura used the Web to conneCt and activate
an army of supporters (Munro, 2000). It is estimated that Ventura’s use of online carn-
paigning contributed to his upset victory by giving him an additional two to four percent
of the vote, which was approximately the size of his margin of victory (Noble, 1999).

Compared to “the meat and potatoes” of presidential campaigning such as tele-
vision advertising, phone books, direct mail and old-fashioned knocking on doors, the
utilization of the lntemet was probably in its infancy in the 2000 presidential campaign
(Zeller, 2000). But the lntemet made a bigstride as a campaign tool in the election: it
was used for a variety of purposes such as recruiting volunteers, soliciting contributions,
spreading candidates’ messages, and even getting voters to the polls.

The John McCain campaign is a good example. He was considered “a
cash-poor underdog” and was anticipated to withdraw early in the race. However, his
online campaigning changed his underdog status to that of a legitimate candidate.
Shogren (2000)e'xplains, “Since McCain’s stunning landslide victory in New Hampshire,
his campaign has raised more than S 2.2 million online, giving him an immediate infu-
sion of cash that is not only keeping his campaign alive but is changing the Arizona Re-
publican's status from that of maverick longshot to serious contender” (p. A20). Fred
Wertheimer, president of Democracy 21, a nonpartisan public policy organization, calls
McCain’s cyber-bonanza “the first big breakthrough example of how the Internet can af-
fect elections in this country” (cited in Shogren, 2000, p. A20).

It was reported that the McCain campaign eventually garnered about 7 million
dollars through lntemet contributions (Munro, 2000). Third party candidates are no

exception. The Nader campaign was reported to have raised about $1 million on the

lntemet (Regan, 2000). It was estimated that a total of $50 million was contributed to
presidential candidates over the lntemet during the 2000 election, even though the
amount was a fraction of the total campaign finance (Regan, 2000).

The implications of lntemet election campaigning are not limited to the candi-
date side. Internet campaigning has changed the form of political information and in-
creased the amount of information available for voters. The most important difference
between online campaigning and traditional media such as mail, telephone and television
is that voters can control access to the lntemet and retrieve what they want at their con-
venience. Put in another way, the lntemet allows voters to get unfiltered information
directly from information sources, e.g., candidate Web sites.

Hockaday and Edlund (1999) say, “Voters use campaign Web sites for a depth of
information that is unavailable in other media” (p. 14). Voters who visit campaign Web
sites tend to look for serious, substantive political information such ,as issue sections,
candidate biographies and comparative sections. According to Hockaday & Edlund,
these kinds of information are what undecided voters use for decision making, which in-
dicates that candidate Web sites influence the decision making for voting. However,
their assertion was made with anecdotal evidence. The influence of campaign Web sites
needs to be empirically explored.

Candidate Web Sites and Descriptions

There is no “industry standard” criterion with which one can evaluate campaign
Web sites. Reavy (1999) suggests several key issues that can be used to evaluate the
efi‘ectiveness of a Web site. These are Content, Audience, Purpose, Interactivity,
Timeliness, Appearance, and Linkage (CAPITAL). Ireland and Nash (1999) list 10 cri-

teria to evaluate a campaign Web site: (1) online credit card contributions, (2) volunteer

sign-up form, (3) form to collect e-mail addresses, (4) form to collect U.S. postal ad-
dresses, (5) links to issues, (6) newsletter sign-up, (7) sign-up for alerts, (8) events cal-
endar, (9) updating and providing date of last update, and (10) a download time of . 15
seconds or less. Although they do not list candidate’s biographical information and is-
sue standing, these topics are also preconditions for effective campaign Web sites.

One of the common points in the above listed checklists is interactivity. The
term interactivity is defined as “the back-and-forth transactions between a Web site and
the person using that site” (Reavy, 1999, p. 250). Why is interactivity important in
online campaigning? First, interactivity is a unique feature that other traditional cam-
paign media do not have. Second, voter interaction with campaigning can generate
“virtual proximity” which can help voters feel that the person with whom they are com-
mtmicating is nearer than he/she actually is (Reavy, 1999). For these reasons, it can be
stated that the more interactive a Web site, the better it will be received by visitors.

Applying media richness theory, Park and Choi (2001) content analyzed the
2000 presidential candidates’ Web sites. The theory posits that the effectiveness of me-
dia is contingent on its “information carrying capacity” (Daft & Lengel, 1984, 1986).
Therefore, by measuring the information carrying capacity, one can evaluate the effec-
tiveness of a Web site. The authors analyzed candidates’ home pages and the first-layer
pages directly linked to the home pages based on the four factors which consist of infor-
mation carrying capacity: (1) interactivity, (2) multiple communication cues, (3) person-
alization, and (4) case of navigation. The results are presented in Table 1. The Gore
Web site was found to be the richest site in all four categories. The author concluded
that the Gore site provided more interactive features and devices, and thus alloWed

visitors to interact with the site more effectively and easily.

Table 1: Media'Richness of the Presidential Candidate Web Sites

 

 

 

 

 

 

 

 

 

 

 

 

Candidate Web Site Bush Gore Nader Total .
Interactivity 41 68 26 1 35
M It“ le
9 ‘,‘ 'p, 180 1584 124 1888
5 Communication Cues

to

0
§ Personalization 143 266 35 444
Ease of Navigation 84 187 78 349
Total 448 21 O5 263 2816

Note:

0 Adapted from Parkand Choi (2001).
. Figures in each cell represent the numbers of Web features such as
interactivity and communication cues.

Although Park and Choi (2001) provided a method to evaluate presidential can-
didate Web sites, it is difficult to precisely assess the effectiveness of a Web site because
(1) there are many other factors which determine a Web page’s effectiveness and (2) the
definitions of the effectiveness depend on the purposes of Web site providers and users.
Evaluations without having solved these potential problems could lead to an erroneous
conclusion. The present study, therefore, avoids evaluating the effectiveness of a Web
site, but briefly describes the content of the Bush, Gore and Nader Web sites by focusing
on the most important fmrction, interactivity.

Web Site Organizations: Each candidate Web site offered menu items that in-
cluded online contribution, volunteer sign-up, email newsletter subscription, detailed bi-
ography, and political issues. The Nader site, however, did not offer a. Span-
ish-language version, whereas the Bush and Gore sites did. The main menu of the Bush
site consisted of “Bush-Cheney 2000,” “News & Info,” “Multimedia,” “Issues,” “Calen-
dar,” “Get Involved,” “Contribute,” “Voter Outreach,” “Youth Zone,” and “Toolbox.”
This pattern is observed in the other sites; however, the Gore site provided a variety of
menu items more than the Bush and Nader sites did. For example, on the Gore site,
there were such menu items as “Speeches,” “For Family,” “Gore Store,” and an inde-
pendent menu for multi-media presentations. Further, the Gore site provided much
more information, such as brief news stories on the home page, than the other sites did.

Contribution: . Each site offered an online contribution menu through which
one can donate money using a credit card. Neither site accepted electronic checks;
however, one can print out a form in the Bush site and send it to the campaign office with
a check. The Bush site had a pop-up window which was designed to solicit contribu-

tions, but neither the Gore site nor the Nader site had this “intrusive” pop-up window.

10

The Nader site put the contribution menu just below the top header and gave it a distinct
graphical button, which indicated the campaign camp regarded contribution as one of the
most important objectives on the site.

Volunteer: In the volunteer menu of the Bush and Gore sites, a prospective
volunteer could choose the state in which he/she lives. Thus, a prospective volunteer
was able to know what he/she can do in that state. On the Bush site, one could sign up
for various activities such as “I want to volunteer” and “I want to be a Bush E-Team
Leader.” As was the case in the contribution menu, the Nader site located the volunteer
menu on the top part of the page with a distinguishable button; however, the other sites
located the volunteer menu on relatively lower parts of the sites without any atten-
tion-drawing graphical buttons.

Personalization: The Bush and Gore sites offered more sophisticated functions
of personalization than the Nader site did. For example, the “Issues” sections on the
Bush and Gore sites allowed visitors to choose information specifically relevant to them.
Further, in the “State” sections, information was tailored to a visitor’s state. Conversely,
the Nader site did not offer personalization functions such as “State.”

URL: URLs, or Web site names, matter. According to Cornfield, “A good
Web site name is memorable, easy to spell and advances the message, ideally before op-
ponents realize there’s an issue or election to define for the public” (1999a, p. 3). One
of the reasons to choose a URL carefully is that it is relatively difficult for Web users to
locate correct sites. One may accidentally log on to a rogue site, attack site, or parody
site that is created to deceive and misinform visitors.

The Bush site’s URL is straightforward: www.georgewbush.com. The Bush

campaign also has a URL without the middle initial “w” www.georgebush.com. Both

11

get you the same official site. The Gore site’s URL is not as straightforward:
www.algore2000.com. For one reason or another, www.algore.com is not his official
site. Similarly, the Nader site’s URL is www.votenader.com. It was reported that any
URL which has something to do with Nader and the presidential election, such as
www.nader2004.org, have already been bought up by a loyal Gore supporter (Regan,

2000).

12

CHAPTER II

THEORETICAL FRAMEWORKS

Selective Exposure

The classic political communication study of Erie Country by Lazarsfeld,
Berelsen and Gaudet (1944) found that people tend to selectively expose themselves to
the media message of their preferred candidate. Since then, selective exposure to po-
litical messages has been confirmed in practically every subsequent research (on political
communication behavior (Chaffee & Miyo, 1983). The selective exposure hypothesis
posits that “audience members prefer supportive rather than discrepant messages, in or-
der to increase confidence that they think, feel, or act in a correct or acceptable manner”
(Atkin, Greenberg, & Korzenny, 1979, p. 5). Researchers, however, have yet to inves-
tigate selective exposure toward supportive messages on the Web. Does this tendency
toward selectivity hold in the lntemet era, an era which is drastically different from the
previous eras of traditional communications technologies such as newspapers, radio and
television? I

Technology can facilitate the extent to which people can expose themselves to
the media. In the Renaissance era, the invention of the printing press brought “a fantas-
tic change” to the world where books were scarcely available (DeFleur & Ball-Rokeach,
1989, p. 22). The technology spread literacy—which was once monopolized by priests,
nobles, and scholars—to ordinary people and profoundly affected hmnan life and history.
In the modern ages, it was television that drastically changed our information environ-

ment. Zillmann and Bryant (1985) explain the impacts of television:

13

The fact that the new communication technology [television] allows the ma-
nipulation of audio-visual environments with enormous ease and provides an
abundance of program choices at all times undoubtedly will have significant
behavioral and social effects...[E]xposure to the environments presented on the
monitor is more engaging and presumably more enjoyable than many alternative,
immediate environments. There are few social circumstances and emotional
conditions that are consistently preferred over the environments on the screen

(p.5).

Television also provided people with great control over the messages they re-
ceive. “By pushing hand-held remote controls, any number of events can be'accessed;
and access can just as readily be abandoned in favor of exposure to other events” (Zill-
mann & Bryant, 1985, p.4). This statement is, however, an exaggeration of television’s
functions when compared to the age of the lntemet. Some would argue that the
video-on-demand (VOD)1 service, which gives the audience more control than the regu-
lar television, is available these days. But the service has not been broadly used and the
contents the service can provide are very limited. Television, hence, is not really inter-
active. Television programs are set and the viewers do not have control over television
content.

On the other hand, the lntemet is interactive. The control over content on the
lntemet is incomparable with that of television and other media. On the lntemet, people
can choose and/or manipulate by customization nearly everything they want to see and
hear. For example, myyahoocom allows customers to change the site’s appearance and

contents in order to match individual interests and needs, such as weather forecast where

 

' VOD is a pay-per-view television service in which the audience can select a program
from a menu and have it delivered instantly to the television set (Grant & Meadows,
2000)

14

they live and the prices of stocks they own. Nearly all information on the Web site can
be filtered at their discretion. .

The Internet has also made it possible for users to reach anywhere in the world
by merely clicking or typing a URL. Physical distance is non-existent in cyberspace.
It takes virtually no physical effort for users to reach any site. The site or information
they select will come to their place, their computer screen. It takes only a click to get to
the candidate site on the lntemet. Additionally, users not only can select content to re-
trieve, but also store it to use at a later time. Exposure to content on the lntemet is in
the users’ hands. The lntemet, more than the other media, help people selectively ex-
pose themselves to what they want to see and hear. On the other hand, in a newspaper
or television news program, 'it may take relatively great effort or luck to reach events of
interest. Therefore, the lntemet may enhance the tendency toward selective exposure
by allowing users to see what they want.

Whereas the tendency toward selective exposure has long been taken for granted,
there is much less unanimity in terms of the extent to which this phenomenon of selective
exposure occurs and the factors that account for it (Chaffee & Miyo, 1983). Under-
standing how and why people have this tendency is a pre-condition to examine selective
exposure in the age of the lntemet. Theoretical frameworks that explain selective ex-
posure are reviewed below.

Partisan/Reinforcement Perspective

The partisan hypothesis posits that partisan predispositions motivate a person to
be selectively exposed to favored political messages to reinforce that predisposition. In
the political communication area, it is often asserted that the prior preference for a can-

didate would lead a person to select messages that would strengthen, not weaken, that

15

preference (Chaffee & Miyo, 1983). In other words, conversion from one, candidate to
another by political messages cannot be expected because the prior preference motivates
people to selectively expose themselves to messages that only reinforce their respective
preference.

This . concept of selective exposure was once synthesized into the often cited
“limited-effect model” which is no longer supported by current scholars. The model
was based on the above mentioned political communication study by Lazarsfeld, Berel-
sen and Gaudet (1944), which found that the mass media did not significantly influence
people’s decision for whom they vote. Klapper (1954, 1963) incorporated the findings
in the study and other bodies of evidence to elaborate the mechanism of the limited effect
model. He claimed that attitudinal predispositions “largely determine the communica-
tions to which the individual is exposed” (1963, p. 67) so that “attitude conversions are. ..
comparatively rare” (1954, p. 308).

Berelson and Steiner (1964) articulated selective exposure firrther by stating,
“People tend to see and hear communications that are favorable or congenial to their pre-
dispositions; they are more likely tosee and hear congenial communications than neutral
or hostile ones. And the more interested they are in the subject, the more likely is such
selective attention” (p. 529). Another underlying explanation of this mechanism is

stated by Merrill and Lowenstein (1971).

So what must the sensible audience member do?... He will build up his own
complex ‘safety’ mechanism for screening incoming information; he will see
less and less that does not agree with his dominant dispositions [selective expo-
sure];. .. he uses propaganda to simply reinforce—not challenge—his basic atti-
tudes and predispositions. .If he did not do this, he would quickly fly into a
million emotional pieces in the face of unverifiable and disharmonic information

16

and opinion that surround him everyday (p. 226—227).

Dennis and Chaffee (1978) used a probability sample of voters representing the
state of Wisconsin to examine a selective exposure effect. Four waves of telephone in-
terviews were conducted with 164 respondents to trace a long-term trend of candidate
evaluation during the fall 1976 presidential election. The findings in the panel study
demonstrated that positive evaluations of one’s preferred candidate/party increased as the
campaign progressed, especially among those who were heavily exposed to the campaign
and the F ord-Carter debates. However, a stronger reinforcing effect was found in inter-
personal discussion than exposure to the campaign via media Thus, the overall find-
ings suggested that while selective exposure tothe campaign through the media had a
reinforcing effect, interpersonal discussion exceeded the mass media in terms of the ef-
fect magnitude.

Chaffee and Miyo (1983) interviewed 501 pairs of adolescents and their parents
during the 1980 presidential campaign year to examine whether partisan predispositions
would motivate to selectively expose oneself . to political campaign messages, which in
turn serve to strengthen those predispositions. The study found that the tendency .to-
ward selective exposure was exhibited among both adolescents and their parents.
However, the tendency was found stronger among adolescents, contrary to their expecta-
tion. They interpreted the difference by stating, “Being comparatively new to politics,
the adolescents respond more to the campaign, and they are considerably less likely than
their parents to pay attention to the campaign communication of the candidate who is
running in opposition to the one they favor” (p. 32). They also compiled a list of the

characteristics of selective exposure they found as follows:

17

(1) Selective exposure occurs mainly among those who are less experienced
and informed about politics—the adolescents.

(2) Selective exposure consists mainly of low attention to the opposition candi-
date, not heightened attention to one ’s favored candidate.

(3) While there is a general positive effect associated with exposure to either
candidate, the greatest change in afi‘ect during the campaign is found for the
person’s favored candidate, not the opponent.

(4) Among the adolescents, the predisposition that appears to generate an indi-
rect effect of reinforcement via selective exposure is identification with a
political party, not specific liking of candidates (p. 32).

Sweeney and Gruber (1984) surveyed voter information preferences during the
Senate Watergate hearings of 1973 in order to examine selective exposure theory. A
three wave panel study was conducted: (Wave 1) just before the Watergate hearings2
started, (Wave 2) midway through the hearings, and (Wave 3) just before the end of the
hearings. The survey identified three types of respondents from a total of 82 interview-
ees who participated in all three surveys—Nixon supporters, McGovern supporters, and
undecideds. It then examined their responses to questions regarding the incident. The
researchers found that the Nixon supporters reported less interest .in and less attention
paid to Watergate-related matters than did members of the other groups. They were also
found to be less likely to engage in Watergate discussion. Further, the Nixon supporters
seemed to have less knowledge about the committee proceedings, perhaps due to selec-
tive exposure (or avoidance), than the undecided or McGovern supporters. Specifically,

they were able to recall fewer names of those involved in Watergate and of the Senators

 

2 The Senate Watergate hearings were set to investigate improprieties surrounding the
conduct of the Nixon reelection committee. However, the hearings eventually focused
on the President’s White House staff and the President himself.

18

who were serving on the committee than were either the undecideds. or the McGovern
supporters. Yet, there is an alternative interpretation that these differences between the
Nixon supporters and the other groups could be attributed to the possibility that Nixon
supporters were generally less interested in and less knowledgeable about politics.
However, the authors found no significant difference in terms of general political interest
and knowledge between the Nixon supporters and McGovern supporters.

As presented before, partisan perspective of selective exposure seems to align
with an online campaign study. According Cornfield (1999b), “Interactive communica-
tion begins when a person wants to learn more about a subject by going online” (p.31).
In other words, online campaign sites have to be selected for exposure by the audience.
Thus, it can be hypothesized that voters who visit a candidate Web site are likely to be
primed for the candidate or have some predispositions to seek information about said
candidate. They may also log on to a campaign site to reinforce or reconfirm their
preference for a candidate.

Schemata Perspective

Even though our information environment is saturated with political information,
most people remain indifferent to, poorly informed about, and very selective of political
information (Fisk & Kinder, 1981). Fisk and Kinder explain selectivity of political in-
formation including conceptions of politicians, policies, and rules, by using the political
schemata model. Schemata are cognitive representations of generic concepts and con-
sist of the attributes that constitute the concept and relationships among the attributes.

According to Fisk and Kinder, schemata provide cognitive economy, which
means that people construct understandings or expectations of the environment effi-

ciently by depending on available and ready-made prior knowledge. In other words,

19

people do not struggle with constructing new understandings at each encounter with new
information because schemata play a role of gatekeeper through which new information
is screened (Fisk, Lau & Smith, 1990; Ossoff & Dalto, 1996). Thus, when it comes to
political information, political understanding depends on prior knowledge that is an or-
ganized and abstracted version of previous experience.

Cognitive Consistency (Dissonance) Perspective

There is a different perspective by which selective exposure can be explained.
One explanation for selective exposure derives from the human nature that seeks cogni-
tive consistency. There are several theories that refer to cognitive consistency, e.g.,
Heider’s balance theory (1946), Osgood and Tannenbaum’s congruity theory (1955),
Newcomh’s strain for symmetry theory (1953) and F estinger’s dissonance theory (1957).
All posit that individuals deliberately seek out information either to reinforce shaken
convictions or confirm those recently acquired attitudes so as to restore equilibrium in
their system of belief. Of these theories, Festinger’s has been often used and/or quoted
to account for selective exposure mechanism (for example, Cotton, 1980; Cotton, 1985;
Chaffee & Miyo, 1983) probably because the theory parsirnoniously explains how selec-
tive exposure occurs.

According to Festinger (1957), there are three hypotheses regarding the mecha-
nism of selective exposure: (1) if a person holds consistent cognition, he/she has no mo-
tivation for selective exposure, (2) if a person’s cognitive state is moderately inconsistent,
he/she will seek for information that would reduce the dissonance and avoid information
that would exacerbate it, and (3) if the level of dissonance is extremely high, the person
will not seek information that would reduce the dissonance but seek information that

would increase the dissonance.

20

Research on this theory generally demonstrated the hypothesized effects, even
though some studies found inconsistent results (for example, Feather, 1962) and a review
by Sears and Freedman (1967) provided contradictory interpretations. It is argued that
the failure of finding effects can be attributed to design deficiencies and other related
problems (for example, Cotton, 1980). For example, studies which failed to demon-
strate selective exposure effects did not incorporate those factors as curiosity and intel-
lectual honesty, which affect selective exposure (Wichlund & Brehm, 1976).

The first empirical evidence for selective exposure was provided by Ehrlich,
Guttman, Schonbach, and Mills (1957)’s field study. Ehrlich and colleagues examined
the consequences of an important decision for selective seeking of confirming informa-
tion. The study found selective exposure among owners of brand new cars: they
tended to read more advertisements of the car they bought than ads of cars they did not
buy. Stronger selective exposure was observed among those who mused over two or
more cars. Conversely, the tendency of selective exposure was much less manifested
among old car owners who had made no recent decision.

Brock and Balloun (1967) measured selective exposure in an experimental set-
ting. Smokers and non-smokers were exposed to messages that either favored or op-
posed smoking. The study found that smokers made more efforts than non-smokers to
hear a message which disputed the link between smoking and lung cancer, while
non-smokers tried harder than smokers to hear a message asserting a causal link between
smoking and lung cancer. Subjects sought consonance-creating messages and avoided
dissonance-creating messages regardless of smokers or non-smokers.

One typical situation that produces dissonance is making a choice between two

equally attractive alternatives (Cotton, 1985). A presidential election in the United

21

States may also be such a case. US. presidential elections have been usually contested
by two candidates from two major political parties, Democrat and Republican. Some
people might have difi'rculty choosing one; some might not have any difficulty. Re-
gardless of whom people have decided to vote for, they are certainly exposed to negative
information about their preferred candidate. If people’s confirmation for candidates is
swayed by opposing views in newspapers and television news programs, dissonance may
occur. Then, one way to reduce the dissonance is through selective exposure to con-
firming information (Festinger, 1957). Voters may selectively seek information that
supports their decision. A candidate’s Web site may be a perfect place to reduce the
dissonance because the site is designed to persuade people how good, competent, and
beneficial the candidate is.

The selective exposure paradigm is explained by these three perspectives: parti-
san/reinforcement, schemata, and cognitive consistency. Although each perspective
bases the mechanism of selective exposure on a different theoretical fi'amework, all of
these three posit that people tend to expose themselves to supportive messages rather
than discrepant messages. The selective exposure paradigm thus provides hypotheses

that are interesting to examine. These hypotheses are:

H1: A person logs on to one’s preferred candidate’s campaign Web site more
often than the other candidate sites.

H2: The stronger the endorsement for a candidate a person has, the more likely
and frequently a person logs on to the candidate Web site.

H3: The more a person logs on to the preferred candidate Web site, the stronger

one’s endorsement for the candidate will become.

22

Voting Behavior

Much of political communication research has found that exposure to media
coverage of elections is likely to increase interest in politics and eventually voting turn-
out or intention to vote (for example, see Atkin, Galloway, & Nayman, 1976; McCombs
& Mullins, 1973, Patterson, 1980; Pinkleton, Austin, & Fortman, 1998). It is also found
that the level of information voters gained via the mass media increased the probability
of voting (for example, see Palfrey & Poole, 1987). This significant role of the mass
media on voting is probably ascribed to the fact that “[t]oday’s presidential campaign is
essentially a mass media campaign” (Patterson, 1980, p. 3). Few voters have an oppor-
tunity for first hand experience of a presidential election campaign; voters cannot help
but rely on mass media coverage for nearly all election information.

As for candidates, they also have to depend on the media to get across their
messages to the electorate. In the past, for example, the nominating process was mostly
controlled by party leaders. Candidates had to be liked by party leaders, or they did not
have a chance to be nominated. . In the modern age, as voters have lost partisan loyalties,
party leaders’ power has weakened. Candidates have to be covered favorably by the
mass media, or they have little chance of being nominated and eventually elected as
president.

Political communication research provided a lot of empirical evidence that mass
media increase (1) voters’ interest in politics and eventually (2) voting turnout or inten-
tion to vote. The first step, stimulating the electorate ’5 political interest is important
because interested people are more likely- to go to the polls (Campbell, et al., 1960; Mil-
brath, 1965). Research findings consistently demonstrated the link between media ex-

posure and audience’ interest in politics and elections.

23

Hypothesizing the amount of exposure to campaign news (newspaper, television,
radio and news magazines) would be functionally related to the level of political knowl-
edge and the degree of campaign interest, Atkin, Galloway and Nayman (1976) exam-
ined the causal flow among mass media exposure, political knowledge and interest.
They conducted a secondary analysis of the 1972 election data and a panel survey on
college students. The data from these two were then combined to draw inferences.
The findings demonstrated that mass media exposure was functionally related to political
knowledge and to political interest, which suggested that media exposure contributed to
enhance political knowledge and interest. The pattern of the findings also indicated that
each factor (media exposure, political knowledge and interest) stimulated increases in the
other, i.e., mutually reciprocal relationships. The authors interpreted these findings by
stating, “[A] person’s basic interest in politics probably leads him to read and watch news
about a particular campaign; in turn, this exposure arouses more exposure behavior.
Similarly, ‘ there may be an exposure-knowledge spiral, such that attention to political
content generates a better informed person, who then seeks to keep up with the events of
the campaign” (p. 237).

While Atkin et al. suggested a rather reciprocal relationship between media
exposure and political interest, McCombs and Mullins (1973) found that media exposure
led to political interest, not vice versa.. Their study was based on the University of
California, Los Angels (UCLA) longitudinal survey data, which was a random
cross-section of the entire student body and surveyed the same individuals through four
or more years of college. A cross-lagged correlation analysis revealed that the correla-
tions between media exposure at Time 1 and political interest at Time 2 were much

higher than the correlations of political interest at Time 1 and media exposure at Time 2.

24

In other words, media exposure enhanced political interest. The authors delved into the
relationships between media exposure, political interest, and preference for media con-
tent. Although reciprocity among these factors could not be eliminated, the findings
suggested that media exposure increased political interest, and then the increased interest
enhanced the preference for public affairs media content.

Patterson (1980) analyzed longitudinal trends of voters’ interest during the 1976
presidential election campaign by focusing on the roles of media as political interest
stirnulators. The panel survey of 1,236 eligible voters found a positive relationship be-
tween media exposure and interest in the election. The relationship was found at every
stage of the campaign—the primaries, the conventions, and the general election; however,
the impact of news exposure was especially strong during the primaries. Although the
reciprocal relationship between election interest and media exposure could not be elimi-
nated, the author concluded that the efi‘ect 6f media exposure was more powerful than
that of prior interest in election. He also compared the interest degree of the early stage
in the 1940 election and that of the 1976 and found higher interest in the 1976 election.
He attributed the increased interest not only to the larger public role in the nominating
process but also to the more intense media coverage with television news, which was
practically non-existent in the 19403 presidential elections. 3

McLead, Glynn and McDonald (1983) tested whether reliance on television
would influence the way people make voting decisions by a panel study with 97 Wiscon-

sin voters in the 1980 presidential election. The panel found that television reliant vot-

ers used candidate image characteristics in making their voting choices more than did the

 

3 There is an alternative interpretation for the difference. Higher interest may be attrib-
utable to the fact that the 1976 election was held after the Watergate affair.

25

newspaper reliant.

The effect of media exposure on voting is not limited to presidential elections.
Shields, Goidel and Tadlock (1995) examined the different impact of media exposure in
the 1988 and 1990 US. Senate and House elections. They based their study on the em-
pirical evidence that senators have less control than representatives over media coverage
on their performance (Abramowitz, 1980); i.e., “[I]n House elections the vast majority of
campaign messages received by an individual voter will favor the incumbent. In Senate
elections the distribution of messages favoring the incumbent and favoring the challenger
should be much more balanced” (Shields, Goidel & Tadlock, 1995, p. 419) because Sen-
ate elections generally draw more attention and are more competitive than House elec-
tions. They investigated if these different levels of control over media coverage had
any impact on voting decision. As predicted, they found that high levels of media ex-
posure tended to be associated with an increased likelihood that respondents would vote
for the House, though not for Senate, incumbents. In addition, media exposure was
found to have the greatest impact on voting among independents.

The roles of the media on voting were empirically tested by incorporating con-
tent analysis of media coverage on elections and in other countries. For example,
Kleinnijenhuis and Oegema (2001) used a 13-wave panel study and content analysis of
the media coverage of television news to examine whether television news during the
1998 Dutch general election campaign had any effect on voters. A total of 516 respon-
dents participated in the panel survey, and 794 news stories broadcast in prime time tele-
vision news programs from the two major television networks were content analyzed and
then sorted by week. The results showed that television news affected the weekly shifts

of voters’ party preference and that favorable news for a party increased the probability

26

of a vote for the party.

There are, of course, some researchers whose research contradicts the media
exposure-political interest-voting behavior sequence, especially when it comes to the ef-
fects of negative campaigns. They believe that reliance on negative media campaigning
contributes to low political knowledge, voter disgust, cynicism and then a lack of politi-
cal participation (for example, Cappella & Jarnieson, 1997). Particularly negative tele-
vision advertising has been blamed for demobilizing voters, creating citizen apathy to-
ward politics and eventually diminishing people’s interest in voting (for example, An-
solabehere & Iyenger, 1995). However, other findings demonstrated that negative mes-
sages could positively contribute to individuals’ interest in politics or made no differ-
ences in voting intention (for example, see Finkel & Geer, 1998; Garrarnone et at., 1990).
The effects of negative political television advertising are beyond the scope of this study;
however, empirical evidence suggests that positive effects of media exposure such. as
television news and newspapers override these negative effects.

Pinkleton, Austin and Fortrnan (1998) examined relationships among mass me-
dia use, political disaffection, political efficacy and participation. They conducted tele-
phone interviews on 582 registered voters in Washington State during the week prior to
the November 1994 election. The findings were contrary to concems about negative
effects of media coverage on political participation. The study demonstrated that mass
media use strongly predicted positive voting behavior and a higher likelihood of voting,
even though it also found that media coverage surely had negative effects on voters by
increasing cynicism about politics.

While well-documented about the roles of the traditional media in voting be-

havior, research findings about newer forms of media such as radio and television talk

27

shows are scarce and less consistent (Weaver, 1996). Pfau and Eveland’s study (1996)
was of a few to investigate the influence of non-traditional news media. This study
used a path analysis of panel data (N = 151) to compare the influence of traditional (tele-
vision news, newspapers and news magazines) and non-traditional (talk and interview
shows) forms of news media in the initial phase of the 1992 presidential election. The
results indicated that whereas non-traditional media exerted greater influence on the per-
ceptions and attitudes during the initial phase, the influence of traditional news media
exerted greater impact on voting intentions in the final phase of the campaign. .

Of course, the roles of Web sites on voting behavior seldom have been docu-
mented because the lntemet is the most recent addition to campaign media, but should be
investigated. Since the characteristics of the lntemet are quite different from the other
media, the roles of political Web sites may be distinct fi'om those of the older media. It
might be true that those who sought political information either through candidate Web
sites and/or news sites are more motivated to vote on election day. However, due to the
non-existence of research on the influence of political Web sites, it is impossible to de-
rive a firm hypothesis. Therefore, this study presents a first research question which

will explore Web sites’ role on voting behavior:

RQl: Is there any relationship between logging on to candidate Web sites (or

news Web sites) and reported voting?

Third Party Supporters
American voters are socialized into a two-party norm, which is constantly rein-

forced by the media portrayal of elections between Democrats and Republicans (Rosen-

28

stone, Behr, & Lazarus, 1996). Third party supporters must often face a lot of difficul-
ties, such as ridicule from friends, and must face up to a reality that their candidate has
no chance of winning. Most importantly, they must make great effort to gather infor-
mation on their third party candidate simply because the mainstream media rarely cover
_ third party candidates.

In the 1980 presidential election, the major newspapers and weekly news maga-
zines gave Reagan and Carter about 10 times more coverage than all 11 third party and
independent candidates combined (Rosenstone, et al., 1996). Television coverage has
the same tendency to neglect third party candidates. For example, televised presidential
debates usually exclude third party candidates with some exceptions, such as Ross Perot
in the 1992 election. Although third party candidates are sporadically covered during
the early stage of elections, they disappear from the media soon after an election is in firll
swing. This is especially the case in the final crucial weeks of an election. In addition,
the media not only ignore third party candidates but also at times are hostile toward them
(Rosenstone, etal., 1996).

Third parties have tried to counterattack the mainstream media’s neglect and
hostility by publishing their own newsletters and newspapers. Yet, unlike television,
radio and commercial newspapers, these publications are read only by faithful supporters
and cannot reach the general public who may become supporters. Media coverage is an
essential component of a successful campaign, thus third party candidates are handi-
capped from the very start of a campaign. But the present formation of the media has
drastically changed with the inception of the lntemet. Only a decade ago, the media
practically meant the mainstream media such as television news and newspapers. Now,

the lntemet, which gives major parties and third parties equal means of communication,

29

has become an important component of the media.

The lntemet can provide ample information. for third party supporters. . Since
third party candidates are not covered by the major news media as much as major party
candidates, third party supporters are expected to seek information from Web sites. As
was the case for voting behavior and lntemet use, there is no research with which a hy-
pothesis could be made. Thus, a second research question is presented to explore the

use of Web sites by third party supporters:

RQ2: Are third party candidate supporters more likely than major party candi-

date supporters to log on to candidate Web sites and news Web sites?

Uses and Gratifications

The selectivity for preferred and supportive political messages would be only
one of various factors that determine the exposure decision. Online campaign sites
should be selected by the audience for a wide variety of uses and gratifications which
might surpass the above mentioned selective exposure models. The selective exposure
paradigm sees the audience in a negative light: it posits that the audience is active only in
the sense of seeking consonant messages and avoiding discrepant ones. On the other
hand, the uses and gratifications paradigm presents a more positive: image of the audi-
ence: it postulates that the audience follows their interest by choosing the media content
according to their personal needs and goals (Katz, Blunrler, & Gurevitch, 1974; McLeod
& Becker, 1981).

Most media effect models or theories including selective exposure assume that

the audience is rather passive. The application of “passive audience” propositions to

30

online campaigning may not be relevant. Information on the Web is not “push”that is,
a passive audience can obtain information without effort, but “pull,” so the audience may
have to do something to obtain it. The audience has to have not only a prior knowledge
but also motivation and some degree of need in order to choose a specific Web site on
which he or she can obtain desired information. Chaffee and Karrihan (1997) explain
that the passive audience model “is not especially appropriate for understanding political
learning, since many people actively seek information” (p. 425).

The uses and gratifications perspective fills the gap because it incorporates what
the selective exposure perspective lacks: audience motivation and behavior for a specific
media use. According to Rubin (1994), “Uses and gratifications is a psychological
communication perspective that shifts the focus of inquiry from the mechanistic perspec-
tive’s interest in direct effects of media on receivers to assessing how people use the me-
dia” (p. 418). Uses and gratifications is a new paradigm that shifted research focus
from “how the media affect the audience” to “what the audience does with the media”
(Rubin, 1994, p. 418). It posits that the audience is active so that media choice lies with
the audience to satisfy needs and desires (Katz, Blumler, & Gurevitch, 1974). The ob-
jectives of uses and gratifications research (Katz et al., 1974) are: to explain how people
use media to gratify their needs; to understand motives for media behavior; and to iden-
tify functions or consequences that follow from needs, motives, and behavior.

Although uses and gratifications theory possesses these advantages, lack of
theoretical rigor is one of the weaknesses of the theory (for example, McLeod & Becker,
1981), and research using uses and - gratifications tended to be unsystematic (Rubin,
1994). Uses and gratifications research, however, has become systematic and diverted

from an audience motivation focus to a synthesis of audience motivation and media cf-

31

fects. This is a logical convergence since both perspectives have a major similarity:
both seek to explain the outcomes of mass communication.

It has been proved that the paradigm of uses and gratifications is applicable to a
variety of mediated communication situations such as psychological motives, communi-
cation channels, and communication content (Lin, 1996). On the other hand, the appli-
cation of uses and gratifications was limited to a static media landscape of print and
broadcasting communication. However, as new communications technologies and de-
vices have been introduced into our communication environment, uses and gratifications
research foci have been diversified and applied to such new communications technolo-
gies as the VCR (Cohen, Levy, & Golden, 1988), the home computer (Perse & Dunn,
1998), and most'recently, the Internet (Korgaonkar & Wolin, 1999; Lin, 1999).

Korgaonkar and Wolin' (1999) studied Web users’ motivations and concerns
from a perspective of business use. The authors first conducted a series of six focus
groups to elicit responses in terms of Web use, and then constructed the survey question-
naire based on the data fi'om the focus groups. Results obtained from a total of 420.
Web users revealed the presence of seven motivations and concerns regarding Web use.
These seven factors in rank order are as follows: (1) social escapism motivation, (2)
transaction-based security concerns, (3) information motivation, (4) interactive control
motivation, (5) socialization motivation, (6) privacy concerns and (7) economic motiva-
tion.

The first factor, social escapism, refers to the Web as entertainment in which it
allows users to escape from reality and it provides diversion and enjoyment. Transac-
tion-based security concerns are about users’ fear of lntemet security problems. Infor-

mation motivation, the third factor, describes how respondents use the Web for their

32

self-education and information needs. Interactive control motivation means Web users
ability to control over content by personalization and customization. Socialization mo-
tivation represents the role of the Web as a facilitator of interpersonal communication and
activities. Privacy concerns refers to privacy in general and are different from the sec-
ond factor, transaction-based security concerns. The last factor, economic motivation,
means users’ motivation to shop wisely and get free products.

In addition, these seven factors, along with age, income, gender and education
levels, are significantly related to the Web use. For example, the findings demonstrated
that heavier Web users were. more likely to appreciate the Web by rating it higher than
light users; that those who experienced “e-shopping” were more likely to accept the pri-
vacy and security issues and enjoy the interactive features of the Web; and that business
Web users were less worried about the privacy issues while they were more appreciative
of the Web because of economic and convenience reasons.

While little research has explored the political uses of the Web, there are nu-
merous studies about the political uses of . the other media such as newspapers and televi-
sion. news. Research consistently found that newspaper use is associated with informa-
tional uses and gratifications including political knowledge (Chaffee & Kanihan, 1997).
Whereas reliance on television news correlated negatively with knowledge of issues,
newspaper use was strongly associated with political knowledge (Culbertson & Stempel,
1986). Atkin (1972) found in an experiment that an announcement of a discussion
about a social event for the next class increased significantly the rate of students’ use of
the newspaper. Kanihan and Chaffee (1996) found that political involvement induced
by a class discussion of current events leads to a significant increase in newspaper read-

ings (cited in Chaffee & Kanihan, 1997).

33

Similarly, past research found that people use newspapers for political informa-
tion necessary to vote for a specific candidate (for review, Chaffee & Kanihan, 1997).
Other studies also found that people who need political information for various reasons
tend to read the newspaper. When it comes to the influence of television, however, no
general conclusion has been reached. Some studies found that television news is not as
informative as newspapers regarding political issues. However, recent studies found
that television news viewing was associated with viewers’ knowledge about candidate
biographies.

As for the political uses and- gratifications of the media, Blumler and McQuail
(1969) developed eight different motivations for use and non-use of the political contents
of the media. They listed five use motivations: vote guidance, reinforcement of deci-
sions, general surveillance of the political environment, excitement and anticipated utility
in future discussion. Three motivations for non-use or avoidance of political messages
are feelings of alienation, partisanship, and avoiding stress.

Becker (1979) refined the measurement of gratifications and clustered gratifica-
tions into five categories: surveillance, vote guidance, reinforcement, communication
utility, and excitement (mentioned in detail in the method section), which are adapted for
the present study. These gratifications categories, however, were constructed exclu-
sively for the old media. Consequently, the features of the Internet were not incorpo-
rated. Intemet’s features, especially interactivity, may be salient gratifications people
seek from online campaigning and should be examined in the research. The third re-

search question is to explore Web gratifications:

RQ3: What gratifications do individuals seek from online campaign Web sites?

34

Past research consistently found that gratifications are not media specific except
for political knowledge gratification (for example, Becker, 1979). Specifically, people
seeking a specific gratification from television seek that gratification from the newspaper.
However, people may seek different gratifications from online campaign sites because, as
stated above, the lntemet is significantly different from other media in terms of interac-
tivity, personalization, and a “pull” function.

Another important question is whether or not online political campaigning (e. g.,
candidate Web sites) can displace or supplement the traditional media in the future. To
answer this question, first it is necessary to understand the extent to which the online
campaign is a functional alternative to the traditional media (Ferguson & Perse, 2000).
If online campaign Web sites are used for the same or similar reasons as those of the tra-
ditional media, i.e., function as alternatives to the traditional media, there is a possibility
that online campaigning will displace or at least supplement the traditional media in the
future.

For example, newspapers once dominated the acquisition of political informa-
tion before the inception of television. Now television is used as the most significant
political information source (Graber, 1980; Patterson, 1980). The reasons that televi-
sion became the primary information source by displacing newspapers are: (1) television
is a functional alternative to newspapers, (2) it has features which newspapers do not,
such as transmission of audio and visual images, (3) it can broadcast live, and (4) it pro-
vides information for free. The lntemet has at least a couple of additional features to
these reasons: users’ control over content and interactivity. These lntemet features are
inherent advantages over television. Thus, it is important to test whether the lntemet is

a functional alternative to television in other gratifications categories, such as surveil-

35

lance and vote guidance. Four research questions are constructed to analyze and com-

pare uses and gratifications between the traditional news media and the online media:

RQ4: Are there any differences in gratifications people seek from Web sites and
other information sources?

RQS: Is the lntemet an alternative to traditional media?

RQ6: Are the uses and gratifications for Web use a predictor for online use?

RQ7: Is there any difference in lntemet uses and gratifications between major

party supporters and third party supporters?

36

 

CHAPTER III

METHOD

Research Design

A two-wave panel survey was conducted between September (Time 1) and No-
vember (Time 2) 2000. Since the same individuals answer questionnaires repeatedly in
a panel design survey, the changes in their thinking and behavior can be traced for a
long-term period. This panel provides an opportunity to explore the relationship be-
tween voter preference, media exposure and voting behavior over time. Time I prefer-
ence and predisposition are used as predictors of Time 2 media exposure patterns and
voting behavior, while controlling for exposure experience variables at Time 1 and other
variables, such as demographic information and partisanship.

Respondents of this study are college students. Data for Time 1 were collected
during the second and third weeks of September 2000 from three major universities, lo-
cated in Michigan, Missouri and Texas. The number of respondents who participated in
the Time 1 survey is 461. The majority of respondents, 355, were from a university in
Michigan. The numbers of respondents fi'om Texas and Missouri were 58 and 48, re-
spectively. All classes in which data were. collected were social science courses, except
the courses used in Missouri were humanities courses. Respondents were asked to par-
ticipate again in the survey after the presidential election in November 2000. The
number of respondents who participated-in the Time 2 survey is 366: 273 from Michigan,
52 from Texas and 41 from Missouri. Class attrition and absences were responsible for
the decrease in the number of participants from Time 1 to Time 2.

The purpose of this research is not to estimate parameters for the general public

37

in order to generalize the results, but to test the hypotheses and answer the research ques-
tions about the implications of the latest technology in the 2000 Presidential Election
campaign. Whereas the results of this study may not be generalized, this study will
contribute to new empirical evidence to the literature. Additionally, despite the re-
stricted representativeness of the respondents, this study has several methodological and
sample advantages. First, panel design allows analysis of time-order relationships
among variables, which is essential for making inferences about cause and effect rela-
tionships. In addition, panel design is more sensitive in detecting spurious relationships
than one-time measurement methods, such as a cross-sectional design.

Second, in terms of lntemet access, the sample has an advantage over the gen-
eral public: all the respondents have access to the lntemet because these three turiversi-
ties provide students with direct Ethernet connections to the lntemet in computer labs
across campuses and/or in dormitory rooms. Although lntemet access has been expo-
nentially spreading among the general population, users of the Internet are still skewed to
younger generations (Morris, 1999). For example, in November-December 2000 while
75 percent of those between ages 18 and 29 had lntemet access, only 15 percent of those
65 and over had access (Pew Internet & American Life Project, 2001). All of the re-
spondents of this study were “wired” to the Internet, at least theoretically. Thus, un-
qualified respondents—without access to the lntemet—were very unlikely to be included
in this survey.

Hypotheses and research questions which the present study tests are as follows:

H1: A person logs on to one’s preferred candidate’s campaign Web site more

often than the other candidate sites.

H2: The stronger the endorsement for a candidate a person has, the more likely

38

and frequently a person logs on to the candidate Web site.

H3: The more a person logs on to the preferred candidate Web site, the stronger
one’s endorsement for the candidate will become.

RQl: Is there any relationship between logging on to candidate Web sites (or
news Web sites) and reported voting?

RQ2: Are third party candidate supporters more likely than major party candi-
date supporters to log on to candidate Web sites and news Web sites?

RQ3: What gratifications do individuals seek fi'om online campaign Web sites?

RQ4: Are there any differences in gratifications people seek from Web sites and
other information sources?

RQ5: Is the lntemet an alternative to traditional media?

RQ6: Are the uses and gratifications for Web use a predictor for online use?

RQ7: Is there any difference in Internet uses and gratifications between major

party supporters and third party supporters?

Questionnaire and Variables

A questionnaire was prepared to retrieve voter preference, exposure to campaign
Web sites, and gratifications from online campaign sites. Time 1 and 2 questionnaires
are provided in Appendix A and Appendix B, respectively. Respondents were asked to
write their names on the questionnaires so that the present investigator would be able to
identify and match Time 1 and 2 questionnaires. Although respondents wrote their
names, confidentiality was stressed by stating, “Your name on the questionnaire will be
transformed to a code name which could not be traced back to you.” Expressing the

confidentiality during the survey administration is expected to reduce socially desirable

39

answers and allow for freer expression of political opinions and preferences. Question-
naire items and corresponding variables are presented as follows (A list of variables and
recoded variables is provided in Appendix C):

Voter Preference/Partisanship Variables

One of the independent variables in the present study is respondents’ candidate
choice. Three questions about voting choice are used to classify respondents. The
first question is “Which candidate would you vote for if the election is held today?” for
Time 1; “Which candidate did you vote for?” for Time 2. Then, respondents are di-
vided into five types: Bush supporters, Gore supporters, Nader supporters, other candi-
date’s supporters, and undecideds (“Did not vote” for Time 2).

The second question in the questionnaires asks respondents about the degree of
their candidate preference. This is measured by a 5-point Likert scale of respondents’
support for the candidate. Respondents mark their degree of support for the preferred
candidate where 1= a little support and .5 = strong support. These variables along with
media exposure variables listed below are used to test Hypotheses 1 to 3 and Research
Questions 1 and 2. The third question asks respondents about their partisanship regard-
less of their candidate preference and official membership. This variable is used for a
control variable in multivariate analysis.

Media Exposure Variables

Measurement of media exposure includes respondents’ use of: candidate Web
sites, news Web sites, television news programs, newspapers, radio news, and news
magazines. In this study, exposure to Web sites is obtained by asking, “Estimate how
often you have visited to each: candidate official Web site during the presidential election

campaign.” This study operationalized exposure to a Web site as the frequency of log-

40

ging on to Web sites regardless of duration. Unlike the traditional media audiences,
Internet users have to choose a Web site either by typing a URL or clicking a hyperlink;
therefore, it can be assumed that they pay attention to Web sites whenever they log on.
Further, the purpose of this study is not to measure the duration of Web surfing, but the
use of specific Web sites, such as candidate Web sites and news Web sites. Conse-
quently, frequency measurement is more appropriate for the purpose than duration meas-
urement.

Whereas Web site use is measured by frequency, the use of non-lntemet media
such as television news and newspapers is measured by duration (“Estimate how much
time you spent watching or reading the following news sources”). Questions about the
media exposure do not accompany a rating scale ranging from “rarely” to “very free
quently” with numeric values such as 1 - 7 or 0 - 6. Because the interpretation of
“rarely” or “frequently” differ significantly fi'om person to person (Sudman, Bradbum, &
Schwarz, 1996), questions about media exposure are open-ended.

Gratifications Variables

Gratifications are measured for each information source, i.e. candidate Web sites,
news Web sites, newspapers, and television news programs. Gratifications items pre-
sent respondents with a list of reasons for use by wording of, for example, “To see how
the candidates stand on the issues, I use information from the following sources.”
These items are designed to comparatively measure the levels of gratifications which re-
spondents seek from these four information sources. Respondents answer the degree of
each reason. Each item ranges from “strongly disagree” to “strongly agree” on a
5-point Likert scale. A higher score indicates the greater reason for watching or reading

a given information source.

41

Items in the list are drawn fi'om uses and gratifications research about political
message/information by Becker (1979) and McLeod, Bybee and Durall (1982). These
items reflect the most salient gratifications for political information and are modified to
fit online campaigns. In addition, the most unique feature of the Internet, interactivity,
is incorporated in the list. The interactive items are derived from a study by Korgaon-
kar and Wolin (1999). The gratifications items are as follows:

0 Surveillance/vote guidance: To see how the candidates stand on the issues; to help
make up my mind how to vote in the election; to see what the candidates would do if
elected; to judge what candidates are like.

0 Reinforcement: to remind me of my candidate’s strong points; to get information
which agrees with my positions.

0 Communication utility: to use what I learn in politics; to give me something to talk
about with other people.

0 Interactivity: because it is interactive; because it gives me the control over what and
when I want to use it; to participate in the Presidential Election campaign.

Control Variables

All of the respondents have lntemet access; however, some students have their
own computers with Internet access and others do not. There might be differences in
terms of selective exposure to and gratifications about Web sites between the haves and
the have-nots. Therefore, computer ownership is included in the questionnaire as a
control variable. In addition, degrees of comfortableness with computers and the Inter-
net are also asked because, for example, lntemet comfortableness might be a factor
Which affects Web site use. Other control variables are demographic information: gen-

der, age, level in a college and major.

42

Analysis Methods

Hypothesis 1, “A person logs on to one’s preferred candidate’s campaign Web
site more often than the other candidate sites,” is examined by two methods. Differ-
ences in terms of logging on frequency between a preferred candidate Web site and the
other candidate Web sites are examined by t-test. Descriptive statistics tests the differ-
ences in terms of logging on probability.

Supporters for each candidate are divided into three subgroups according to the
degrees of support at Time 1 in order to test Hypothesis 2, “The stronger the endorsement
for a candidate a person has, the more likely and frequently a person logs on to the can-
didate Web site.” Then, AN OVA tests the difference in terms of logging on frequency
among these subgroups, while cross tab analysis with gamma (Goodman-Kruskal’s index
of relationship) is used to examine the differences in terms of logging on probability.

Hypothesis 3, “The more a person logs on to the preferred candidate Web site,
the stronger one’s endorsement for the candidate will become,” is tested by two methods.
First, correlation analysis assesses the association between the degree of respondents’
Tune 2 support for their candidates and the frequency of logging on to the candidates’
Web sites. Since this correlation analysis does not incorporate a longitudinal trend of
support degree, the finding may be spurious. Therefore, the second method employs the
difference between Time 1 support and Time 2 support as a dependent variable in corre-
lation analysis.

Research Question 1, “Is there any relationship between logging on to candidate
Web sites (or news Web sites) and reported voting?” is first analyzed by cross tab analy-
sis to assess the likelihood of voting by logging on to candidate Web sites and news Web

sites. Then, logistic regression analysis measures the odds ratio and probability of vot-

43

ing by incorporating Web site use, other media consumption- and demographic informa-
tion.

Research Question 2, “Are third party candidate supporters more likely than
major party candidate supporters to log on to candidate Web sites and news Web sites?”
is examined by: (1) t-test for the differences in terms of logging on frequency, and (2)
cross tab for the differences in the likelihood of logging on between third party support-
ers and major party supporters.

Research Question 3, “What gratifications do individuals seek fi'om online
campaign Web sites?” and Research Question 4 “Are there any differences in gratifica-
tions people seek from Web sites and other information sources?” are answered by de-
scriptive and comparative analysis of gratifications items.

A principle component analysis and factor analysis construct gratifications in-
dexes for each medium (candidate Web sites, news Web sites, newspapers, and television
news) to prepare for the analysis of Research Question 5, “Is the lntemet an alternative to
traditional media?” Then, the differences between lntemet indexes and the other media
indexes are examined by t-test.

Research Question 6, “Are the uses and gratifications for Web use a predictor
for online use?” is answered by a correlation method which analyzes the associations
between gratifications index for each medium and respondents’ consumption of corre-
sponding medium.

Lastly, the difference in each gratifications index between major party support-
ers and third party supporters is accessed by t-test to explore Research Question 7, “Is
there any difference in lntemet uses and gratifications between major party supporters

and third party supporters?”

44

CHAPTER IV

RESULTS

Participants
‘ Findings reported here are based on 325 respondents who participated both at

Time. 1 and Time 2 surveys. The sample breakdown was 55.4% male respondents (n =
180) and 44.6% females (n = 145). The majority of respondents (n = 167, 51.4%) are
fi'eshmen, followed by senior students (n = 99, 30.5%). Juniors ranked third (n = 39,
12%) and sophomores fourth (n = 14, 4.3%). The rest are graduate students and other
students, e.g., life long education students (n = 6, 1.8%). All respondents were over the
age of 18 and eligible to vote at the presidential election 2000.
Information Media Use

Respondents, on average, spent 2.27 hours a week on television news watching
(SD = 2.75), 1.56 hours (SD = 1.73) on newspaper reading, .86 hours (SD = 2.10) on
lntemet news sites, .44 hours (SD = .87) on news magazine reading, and .41 hours (SD
= .94) on radio news. One third of respondents logged on to candidate Web sites (n =
101, 31.1%). Excluding non-users, candidate Web site users logged on to the sites an
average of 5.27 times (SD = 5.54) during the presidential election campaign. The ma-
jority of respondents used news Web sites (n = 177, 54.5%) to obtain election informa-
tion. The news Web site users visited the sites an average of 10.82 times (SD = 16.99).

On. average, they went online 10 hours a week (range = 0 to 55, SD = 9.21).
Eighty-five percent of respondents said they owned a computer with lntemet access;
however all of the three universities which participants attend provide students with

lntemet access so that the access rate is 100%. According to Pew lntemet Project, 58

45

percent of adult Americans and 75 percent of an age group from 18 to 29 had lntemet
access as of November-December 2001 (Pew Internet & American Life Project, 2001).
Participants’ access rate is much greater than that of other demographic groups. I

The ownership of computers with lntemet access is a factor influencing lntemet
time spending: while those who owned a computer spent 11 hours online per week, those
who do not own spent four and a half hours. The difference is statistically significant (t
= 4.71, df= 3 19, p < .001). This finding suggests that the ownership of computers with
lntemet access is a predictor for longer lntemet spending.
Voting

At Time 1 (September 2000), 43.1% of respondents (n = 140) intended to vote
for Bush, while 37.8% (n = 123) intended to vote for Gore. Nader attracted 17 respon-
dents (5.2%) who intended to vote for him at that time. Five respondents (1.5%) in-
tended to vote for other candidates. The rest of respondents (n = 40, 12.3%) did not in-
tend to vote for any of them. According to the Time 2 survey, which was administered
a week after the presidential election 2000, the reported turnout rate for respondents was
69.1%. This turnout rate is greater than the national average, about 51% (Center for
Voter and Democracy, 2001). Of all respondents, 32.9% (n = 107) reported that they
voted for Gore, 28.6% (n = 93) for Bush, and 6.8% (n = 22) for Nader. Only two re-

spondents voted for other candidates.

Selective Exposure
Hypothesis I predicted that respondents would log on to their preferred
(Time 1) candidate Web sites more often than other candidate Web sites. Charts 1

through 3 and Table 2 generally demonstrate the tendency for selectivity throughout the

46

Chart 1: Bush Supporters Candidate Web Site Use

 

 

 

l
l
l
l
l

 

 

A
.h.
l

 

.3
N
7..

 

A
.._.._JL.

 

Frequency
.0 .0 .0
O) on

.5

 

 

 

 

 

 

   

 

 

Bush Site Gore Site Nader Site

47

Frequency

Chart 2: Gore Supporters' Candidate Web Site Use

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. I:

Bush Site Gore Site Nader Site

 

48

1.6
1.4

Frequency
o .1.
on a» is)

P
or

0.4 .

 

Chart 3: Nader Supporters' Candidate Web Site Use

 

F

l

 

 

 

P,

 

 

 

 

 

 

 

 

 

 

 

 

Bush Site Gore Site Nader Site

49

 

Table 2: Mean Frequencies of Visiting Each Candidate Web Site by Bush, Gore
and NaderSupporters

 

 

 

 

Supporter Bush Site Gore Site Nader Site
Bush Supporters f .86 .45 .14
SD 2.53 1.46 .72
Gore Supporters f .52 1.27 .22
SD 1.31 3.11 1.22
Nader Supporters f .59 .94 1.47
SD 1.18 1.20 2.60

 

 

50

presidential election campaign. Bush supporters visited the Bush Web site an average
of .86 times (SD = 2.53) while they logged on to the Gore site .45 times (SD = 1.46) and
the Nader site .14 times (SD = .72) during the presidential election campaign. The dif-
ferences are statistically significant (mean difference between the Bush site and the Gore
site, t = 2.45, df= 139, p < .05; the Bush site and the Nader site, t = 3.45, df= 139, p
< .01 ). Gore supporters used the Gore site an average of 1.27 times while they logged
on to the Bush site .52 times and the Nader site .22 times. The differences are also sta-
tistically significant (mean difference between the Gore site and the Bush site: t = 3.48,
df= 122,p < .01; the Gore site and the Nader site: t = 3.58, df= 122,p < .001).

Nader supporters visited the Nader site an average of 1.47 times compared to .94
times for Gore’s and .59 times for Bush’s sites. However, the differences are not or are
only marginally significant (mean difference between the Nader site and the Gore site: t
= .91, df= 16, p = .38; the Nader site and the Bush site: t = 1.79, df= 16, p = .09) proba-
bly due to the small sample size of Nader supporters (n = 17).

This tendency of selective exposure was further confirmed, though not outright,
by the probability that respondents logged on to their preferred candidate Web sites.
Table 3 generally demonstrates that respondents were more likely to visit their favored
(Time 1) candidate Web sites rather than the other candidate Web sites. Twenty-five
percent of Bush supporters logged on to the Bush site compared to 17.1% to the Gore site
. and 4.3% to the Nader site. For Gore supporters, 34.1% of them visited the Gore site
while 23.6% and 7.3% visited the Bush site and the Nader site, respectively. Nader
supporters visited the Nader site and the Gore site relatively equally (47.1 % and 41.2 %,
respectively); however, a smaller segment of the Nader supporters (29.4 %) logged on to

the Bush site.

51

Table 3: Percentages of Visiting Each Candidate Site by Bush, Gore and Nader

 

 

 

 

Supporters
Supporter Bush Site Gore Site Nader Site
Bush Supporters 25.0% 17.1% 4.3%
Gore Supporters 23.6% 34.1% 7.3%
Nader Supporters 29.4% 41.2% 47.1%

 

 

52

In sum, although there are a few fluctuations in the findings, the tendency of se-
lective exposure to preferred candidate Web sites was generally observed in the patterns
of respondents’ candidate Web site use. Hypothesis 1 was generally supported. The
finding also verified the effectiveness of an lntemet function, users’ control over the
content, which allows Web users to select what they want to see and read.

Hypothesis 2 posits that the more strongly a person endorses a certain can-
didate, the more likely he or she is to log on to the candidate’s Web site. To test this
hypothesis, supporters of each candidate were first divided into three subgroups accord-
ing to the degrees of support at Time 1: weak supporters, middle supporters, and strong
supporters. Weak supporters are defined as those whose support degrees for the candi-
date were 1 and 2 in 5-point Likert scale. Support degrees of 4 and 5 are categorized as
strong supporters. The rest, support degree of 3, was defined as middle supporters.
The breakdowns of subgroups are as follows: Bush supporters (weak = 31, middle = 52,
and strong = 57), Gore supporters (weak = 22, middle = 47, and strong = 54), and Nader
supporters (weak = 5, middle = 7, and strong = 5).

By using the three types of subgroups according to the degrees of support, the
differences in terms of logging on frequency and probability among the subgroups were
examined. Overall, strong supporters were more frequently and more likely to log on to
their preferred candidate Web sites than weak and middle supporters were. Charts 4
through 6 and Table 4 show the frequencies of visiting preferred candidate Web sites by
the subgroups. Strong Bush supporters logged on to the Bush site an average of 1.67
times compared to middle supporters’ .25 times and weak supporters’ .39 times (F =
5.257, df = 137, p < .01). Strong Gore supporters visited the Gore site an average of

2.24 times whereas middle and weak supporters did so .60 times and .32 times, respec-

53

F requency
3.

Chart 4: Bush Supporters' Logging on Frequency By Support

Degree

 

 

 

 

 

 

 

 

F

 

Weak Supporter

Middle Supporter

54

Strong Supporter

 

Chart 5: Gore Supporters' Logging on Frequency by Support
Degree

 

 

 

Frequency
3

 

 

 

 

 

Weak Supporter Middle Supporter Strong Supporter

55

- Chart 6: Nader Supporters' Logging on Frequency by Support

Degree

 

A

/

 

Frequency
:5

/

 

 

,/

 

 

 

 

Weak Supporter

Middle, Supporter

56

Strong Supporter

 

Table 4: Logging on Frequencies of Each Candidate’s Supporters by Weak,
Middle and Strong Support Degrees

 

 

 

 

Weak Middle Strong
.39 (1.17) .25 (.62) 1.67 (3.70)
Bush Supporters f (SD)
F= 5.26, df= 137, p < .01
.32 (.89) .60 (1.61) 2.24 (4.23)
Gore Supporters f (80)
F = 5.09, df= 120, p < .01
1.00 (2.24) , 1.00 (1.29) 2.60 (2.19)
Nader Supporters f (SD)
F =1.29,df=14,p = .31

 

 

57

tively (F = 5.093, df= 120, p < .01). Strong Nader supporters (2.6 times) logged on to
the Nader site more than twice as many times as middle and weak supporters did (1.00
times for each). However, the difference among Nader subgroups was not statistically
significant (F = 1.29, df= 14, ns.).

In terms of probability, gamma (Goodman-Kruskal’s index of relationship) was
used to measure the strength and direction of the cross-tabular association between log-
ging on probability (yes and no) and the three subgroups (strong, middle and weak sup-
porters). The sign of gamma indicates whether the overall association is positive or
negative, and the magnitude of gamma is the strength of the association. If respondents
are overall higher on one variable as well as higher on the other variable, then the asso-
ciation is positive. On the other hand, if higher on one variable and lower on the other
variable, the association is negative (see, Agresti & Finlay, 1997).

Table 5 demonstrates the positive relationship between the degree of support and
the probability (of logging on to the preferred candidate Web sites: as the degrees of sup-
port increase, the probability of visiting the preferred candidate Web sites goes up.
Specifically, while 16.1 percent of weak Bush supporters logged on to the candidate site,
36.8 percent of strong Bush supporters did so (y = .401, p < .05). Likewise, whereas
18.2 percent of weak Gore supporters logged on to the Gore site, 48.1 percent of strong
supporters did so (y = .461, p < .01). Strong Nader supporters (80%) were four times
more likely to visit the Nader site than weak supporters were (20%). The association
was significant ( y = .69, p < .05) even though the sample size was small.

Cross-tabular analysis may lose some information because the data are collapsed
into subgroups. Thus, the 5-point degree of support at Time 1 for each favored candi-

date and the frequencies of logging on to the preferred candidate Web site were analyzed

58

Table 5: Logging on Probabilities for Weak, Middle and Strong Supporters by Each

 

 

 

 

Candidate Supporters

Sub-groups by -

support degree Weak Middle Strong
tn Logged on 16.1% 17.3% 36.8%
5 (n) (5) (9) (21)
I
8
Q Did not log 83.9% 82.7% 63.2%
g (I?) (26) (43) (36)
_'
m
3° y =.4o1,p<.05
G) Logged on 18.2% 25.5% 48.1%
g (H) (4) (12) (26)
m
‘6
'9 . ‘ 81.8% 74.5% 51.9%
(:33 Did not log (18) (35) (28)
-l
m .
1‘ y=.461,p<.01
g Logged on 20.0% 42.9% 80.0%
g (I?) (1) (3) (4)
:0
‘8
'0 Did not log 80.0% 57.1% 20.0%
3 (n) (4) (4) (1)
1°.
m
23 y = .692, p < .05

 

 

59

by Pearson’s correlation. Generally, the degree of Time 1 support is positively related
with the frequency of visiting the preferred candidate Web sites. For Bush supporters (n
= 140), the relationship between the degree of support and logging on frequency is posi-
tively related (r = .23, p < .01). Similarly, Gore supporters’ endorsement degree is pos-
tively associated with the frequency of visiting the Gore site (r = .25, p < .01). The
correlation coefficient for Nader supporters (r = .47) is stronger than those for Bush and
Gore supporters; however it is marginally significant (p = .06), again probably due to the
small sample size (n = 17).

To sum, respondents logged on to their preferred candidate Web sites according
to not only whether or not they endorsed the candidates but also how much they endorsed
the candidates. The degrees of prior endorsement, or predisposition, determine the de-
grees of selective exposure to their favored candidate Web sites. Hypothesis 2 was sup-
ported.

Hypothesis 3 predicted that selective exposure to a preferred candidate Web
site during the election campaign would eventually enhance the degree of support
for the candidate at Time 2. While the findings above confirmed that the degrees of
endorsement determine the degrees of selective exposure to the preferred candidate Web
sites, the proposition of Hypothesis 3 predicted the causal effects of selective exposure
on the endorsement for candidates. Only those who supported (Time 1) and voted for
(T irne 2) the same candidates were included in this analysis. Thus, converts who sup-
ported a candidate at Time 1 but voted for another candidate were excluded from this
analysis. A total of 224 respondents who qualified for this criterion were identified.
This hypothesis was tested by two methods.

First, correlation analysis assessed the association between the frequency of log-

60

ging on to the candidates’ Web sites and the degree of respondents’ Time 2 support for
their candidates. Overall, the frequency of visiting preferred candidate Web sites and
the degrees of Time 2 support for the candidates are positively correlated (r = .28, p
< .001). The correlation coefficients between Time 2 support and logging on frequency
are: for Bush voters, .27 (p < .05), for Gore voters, .28 (p < .01), and for Nader vot-
ers, .50 (p < .05).

These positive relationships between logging on frequency and support degree
do not necessarily mean that selective exposure to preferred candidate Web sites boosts
the degree of support for the candidates. It is quite possible to put an alternative inter-
pretation on that finding: a strong endorsement for a candidate led respondents to log on
to the candidate Web site and not vice versa, as the findings in the previous section dem-
onstrated. Therefore, a second method was employed in. order to invalidate the altema-
tive interpretation. This method used the difference between Time 1 support and Time
2 support as the dependent variable. The degree of Time 1 support was subtracted from
that of Time 2 support. The value could be positive (the support increased from Time
1) or negative (the support decreased from Time 1), or zero (the degree of support did not
change from Time 1 to Time 2).

Overall, correlation coefficient analysis revealed no significant relationships
between support increase/decrease and the frequency of logging on to the candidate Web
sites (r = -.01, ns). For Nader voters, on the other hand, the support increase/decrease
and the frequency of logging on appear to be positively related (r = .53). However, the
association was not statistically significant (p = .17) due to the small number of Nader
voters who qualified for this analysis (n = 8).

Although respondents who visited their preferred candidate Web sites tended to

61

endorse the candidates more strongly than those who visited less, this analysis could not
confirm that the exposure to preferred candidate Web sites strengthened respondents’
endorsement for the candidates. The analysis observed a weak indication of the rein-
forcement effect among Nader voters; however, causal inferences cannot be drawn from
the finding. As a result, Hypothesis 3 was not supported.

Voting behavior

The first research question explored the relationships between voting be-
havior and the use of candidate Web sites and news Web sites. First, non-parametric
statistic analysis was used to assess the likelihood of voting by logging on to candidate
Web sites and news Web sites. The results are shown in Table 6. Those who logged
on to candidate Web sites were more likely to vote than those who did not (x 2 = 13.54,
df= 1, p < .001). Specifically, 83.2% of respondents who visited one or more candidate
Web sites reported that they voted on election day, while 62.8% of non-visitors did so.
Similarly, those who used news Web sites for the presidential election information were
more likely to vote than others (x 2 = 14.01, df= 1, p < .001). Specifically, 77.8% of
news Web users reported that they voted; 58.5% of non-users did so. Generally, users
of candidate Web sites and news Web sites were more likely to vote than non-users.

The above analysis does not provide how much the probability of voting in-
creased by a unit increase in the frequencies of candidate Web site use and news Web site
use. In addition, the above results might be spurious because the analysis did not in-
corporate such variables as demographic information and other media consumption
which have been found to influence voting behavior. Therefore, multiple logistic re-
gression was used to measure the effect size of these Web sites and to improve the valid-

ity of analysis. Logistic regression quantifies the odds ratio of a dichotomous response.

62

Table 6: Voting Behavior by Online lnforrnation Use

 

 

 

 

 

Did you vote? YES NO
0 Logged 82.3% 16.8%
g (n) (84) (17)
9
D
g 2: Did not Log 62.8% 37.2%%
g 3“, (n) (140) (83)
rn :4
'z' m 2_ _ a
E") x -12.54,df—1,p<.001
m
g Logged 77.8% 22.2%
(23 z (n) (137) (39)
= E
(D
m 3 Did Not Log 58.5% 41.5%
a (n) (86) (61)

x2=14.01,df=1,p<.001

 

63

The dependent variable of this research question is a binary response variable, whether or
not respondents voted for a presidential candidate (coded 1 for yes, 0 for no). Thus, this
statistical method can measure relative contributions of independent variables to the odds
ratio of voting behavior which can be easily converted into the probability of voting.

The primary independent variables are the frequencies of logging on to candi-
date Web sites and news Web sites. In addition, the amounts of newspaper reading,
television news watching, radio news listening, and news magazine reading were entered.
All these variables are continuous variables obtained by open-ended questionnaire items.
Partisanship (whether or not you are oflicially or unofficially affiliated with a political
party: coded 1 for yes, 0 for no), Gender (coded 1 for male, 0 for female) and Level (1=
freshmen to 5 = graduate students) were also entered in the equation because these
demographic elements may influence voting behavior.

A preliminary analysis found an interesting result: all media variables except for
online information sources (candidate Web sites and news Web sites) were not significant
predictors for voting. Respondents’ gratifications for online information sources are
ranked only after newspaper and television news programs (This will be discussed fur-
ther in the uses and gratifications section). Even though respondents generally an-
swered that they did not appreciate online information sources as much as they did the
traditional media such as television news and newspapers, the use of online information
sources are significant predictors for respondents’ higher probabilities of voting. Online
information may have a tremendous impact on voting turnout.

The analysis also found that demographic variables were not significant predic-
tors for voting behavior. Thus, these three variables (Gender, Level and Partisanship)

were eliminated along with non-online media use variables from the equation. The

64

multiple logistic regression was then calculated by incorporating the frequencies of can-
didate Web site use and news Web site use. Table 7 shows the results of the multiple
logistic regression analysis.

The frequencies of visiting both candidate Web sites and news Web sites are
predictors for voting (candidate Web sites: 13 = .1795, p < .05; news Web sites: 13 = .0795,
p < .01). Respondents who visited either candidate Web sites or news Web sites were
more likely to vote than those who visited less or did not visit. The equation which

predicts the log odds of respondents’ voting is shown in Equation 1:
logit (p) = .3598 + .1795 x) + .0795 x2 (Equation 1)

Where x1 is the frequency of candidate Web site use, and x2 is the frequency of

news Web site use.

Coefficient (13) represents the change in the log odds of voting for a unit change
(frequency) in each independent variable (candidate Web site use and news Web site use).
The odds ratios for candidate Web site use and news Web site use are 1.20 and 1.08, re-
spectively. The odds ratios mean that for each unit increase in each variable, the odds
of p change to the product of odds ratio and its previous value. Specifically, the fre-
quencies of candidate Web site use and news Web site use increase by 1, the odds of p
change to 120% and 108% of their previous values, an increase of 20% and 8%, respec-
tively. Odds can be obtained in the same way that odds ratio can be obtained, by taking

the inverse of the natural log of logit (p) as Equation 2 denotes:

65

Table 7: Logit Regression Analysis about Factors Determining Voting Behavior

 

 

 

Voting Behavior (1 = Yes, 0 = No)
Coefficient SE Odds Ratio ell
Online Information Use
Candidate Web Site .1795” .0891 1.20
News Web site .0795*** .0306 1.08
Logit Constant .3598" .1452

 

 

Note:

0 Online use variables, candidate web site and news web site, are frequencies
of visiting each web site. '

- Dependent variable (Did you vote?) is coded as 1 = yes, 0 = no.

0 Significancefor Multiple logistic regression: * p < .10, ** p < .05, *** p < .01

66

Voting Odds = e(.3598 + .1795Xl + .0795X2) (Equation 2)

Where e is the natural logarithm base, x. is the frequency of candidate Web site

use, and x; is the frequency of news Web site use

For example, odds for those who did not log on to either candidate Web sites or
news Web sites are 1.433. Odds for those who logged on to candidate Web sites 10
times, but did not visit news Web sites is 8.63. On the other hand, odds for respondents
who did not visit candidate Web sites, but logged on to news Web sites 10 times is 3.17.
Chart 7 shows that candidate Web sites and news Web sites have the different impacts of
a unit change on odds of voting.

The predicted probability of voting by the frequencies of visiting candidate Web

sites and news Web sites can be obtained by Equation 3:

Voting Probability = odds divided by (odds + 1), or

e(.3598+.1795x, +.0795x2) , 3
1+e(.3598+.l795x, +.0795x2) (Equation )

Where e is the natural logarithm base, x. is the frequency of candidate Web site

use, and x2 is the frequency of news Web site use

Chart 8 demonstrates the difference in terms of probability increase between

candidate Web site use and news Web site use. For example, the probability of voting

67

Tie; ozoz .1 no>> Sage—emu ILL."

mv

 

55:09... om: oc=c0

cv mm on mm cu m _. 0.. m

 

 

om: coszoE. 95:0 3 coocoesm mm mczo> .2 «.25 K :20

 

ON

0v

om

om

. sppo 5U110A

cow

GNP

o:

68

 

 

3.25 952 1...... no>> mamuficmoi

 

om mv ov mm on mm ow

55:09“. cm: 05:5

9.

or m

 

 

 

 

 

on: 8.55.8... 2:5 a 8283 an 88> o 82589.. a :20

 

 

md

". ‘0.
o 0

9°. .
o

Ariiiqeqma 6U110A

ad

69

of non-Web users—those who did not log on to either candidate Web sites or news Web
sites—are 59%. The probability of voting of those who logged on to candidate Web
sites 10 times, but did not visit news Web sites is 90%. On the other hand, the probabil-
ity for respondents who did not visit candidate Web sites, but logged on to news Web
sites 10 times is 76%.

On average, respondents logged on to candidate Web sites and news Web sites
1.64 times and 5.91 times, respectively. For the average online users, the odds of voting
is 3.08 and the probability of voting is 75%. The average respondents’ probability of
voting is 16% greater than those who did not use either candidate Web sites and news
Web sites. This result indicates that exposru'e to online information sources about the
presidential election motivates respondents to vote and the effect of candidate Web sites
on voting is stronger than that of news Web sites. However, the reverse could be true;

respondents who decided to vote were more motivated to use online information sources.

Third Party Supporters

Research Question 2 explored whether third party supporters would use
candidate Web sites and News Web sites «more than major party supporters. In-
formation about third party candidates is scarce in the major media so that third party
supporters may have to rely on Web sites on which they can obtain information they want.
Since there are only a couple of third party supporters who endorsed presidential candi-
dates other than Nader, third party supporters in this study are defined as Nader support-
ers. In addition, this study differentiates supporters from voters: Supporters here are
defined as those who endorsed a certain candidate at the Time 1 survey (September

2000), while voters are defined as those who reported that they actually voted for a can-

70

didate at the Time 2 survey (November 2000).

Table 8 shows the results regarding third party supporters/voters’ use of candi-
date Web sites. While 64.7% of third party supporters at Time I logged on to candidate
Web sites, 31.2% of major party supporters‘did so (x 2 = 8.09, df= 1, p < .01). As for
voters at Time 2, 68.2% of third party voters used candidate Web sites while 34.0% of
major party voters did so (x 2 = 9.89, df = 1, p < .01). The results demonstrate that
third party supporters/voters were more likely to use candidate Web sites.

On the other hand, the third party supporters/voters’ dependency on candidate
Web sites did not hold in their use of news Web sites, as Table 9 demonstrates. The ra-
tios of the use of news Web sites for third party and major party supporters/voters are so
close that the null hypothesis could not be rejected. Specifically, 64.7% of Nader sup-
porters and 58.0% of major party supporters at Time I logged on to news Web sites to
obtain campaign information (7: 2 = .29, df = 1, ns). Similarly, 77.3% of third party
voters and 60.3% of major party voters went online to obtain campaign information
through news Web sites. This difference appears to be large; however, it is not statisti-
cally significant ( x 2 = 2.42, df= 1, p = .12)‘.

In terms of frequencies of logging on to candidate Web sites, descriptive, statis-
tics generally reveals that third party supporters/voters visited candidate Web sites more
often than major party supporters/voters did. For example, third party supporters visited
candidate Web sites an average of 3 times (SD = 3.37) while major party supporters did
so 1.75 times (SD = 4.22); Bush supporters 1.50 times (SD = 4.01) and Gore supporters
2.03 times (SD = 4.44). Third party voters logged on to candidate Web sites an average
of 3.59 (SD = 4.54) compared to major party voters’ 2.02 times (SD = 4.57); Bush voters

1.78 times (SD = 4.30) and Gore voters 2.22 times (SD = 4.79). However, these differ-

71

Table 8: Probability of Using Candidate Web Sites by Major party candidate
SupportersNoters and Third Party SupportersNoters

 

Support (Time1) Major Party Supporters Third Party Supporters

 

3.1.18 83M EIVOIGNVO

 

 

 

Logged 31 .2% 64.7%
(n) (82) (11)
Did not Log 68.8% 35.3%
(n) (181) (6)
x? = .29, df= 1, ns
Voted (TimeZ) Major Party Voters Third Party Voters
Logged 34.0% 68.2%
(n) (68) (15)
Did Not Log 66.0% 31.8%
(n) (132) (7)

x2=9.89,df=1,p<.

 

72

Table 9: Probability of Using News Web Sites by Major party candidate
SupportersNoters and Third Party Supporters/Voters

 

Support (Trme1) Major Party Supporters Third Party Supporters

 

3.1.18 83M SMBN

Logged 58.0% 64.7%
(n) (152) (11)

Did not Log 42.0% 35.3%
(I?) (110) (6)

12:809. df=1,p<.01

 

 

Voted (Time2) Major Party Voters Third Party Voters
Logged 60.3% _ 77.3%
(n) (120) (17)
Did Not Log - 39.7% 22.7%

(n) (79) (5)

12:242. df=1,p=.12

 

 

73

ences were not confirmed by t—test. These insignificant results could be at least partly
ascribed to the wide range of standard deviations, in that the distributions of frequencies
are spread out.

In terms of the use of news Web sites, descriptive statistics also demonstrates
that third party supporters/voters visited news Web sites more often than major party
supporters/voters did. Time 1 third party supporters logged on to news Web sites. an
average of 9.29 times (SD = 12.28) compared to major party supporters’ 6.34 times (SD
= 14.63): Bush supporters 5.94 times (SD = 14.60) and Gore supporters 6.80 times (SD =
14.71). Time 2 third party voters logged on to news Web sites an average of 10.23
times (SD = 21.32) compared to major party voters’ 7.53 times (SD = 15.17): Bush vot-
ers 7.52 times (SD = 16.78) and Gore voters 7.54 times (SD = 13.71). However, all
these difference were not statistically different due to the same reason in the use of can-
didate Web sites.

Overall, third party endorsers were more likely to use candidate Web sites than
major party endorsers whereas the frequencies of visiting were not significantly diflerent
between third party endorsers and major party endorsers. In terms of news Web site use,
on the other hand, there is no difference at least statistically between third party endorsers
and major party endorsers. Bringing together these two findings suggests that a differ-
ence exists between third party endorses’ use of candidate Web sites and their use of

news Web sites.

Uses and Gratifications
Research Questions 3 and 4 explored gratifications which respondents

sought from the online media and examined differences (or similarities) in terms of

74

, gratifications between the online media and the regular media. A total of 11 grati-
fications for each medium, their means and standard deviations are presented in Table 10.
The table demonstrates that there are few gratifications items which are distinct from
other items within each medium. It is especially the case in candidate Web sites and
news Web sites: the mean values for the items are quite similar. This result suggests
that respondents did not seek any specific gratifications fi'om these Web sites, rather they
regarded each gratifications item as similar within each medium.

The table also demonstrates that the mean values of gratifications items in news
Web sites and candidate Web sites are consistently ranked third and fourth, respectively.
Every item in candidate Web sites was rated the least. Television news programs were
ranked as the most gratifying in their utility for a presidential election information source
followed by newspapers except for one item, i.e., control over the content. This result
suggests that these traditional media such as television news programs and newspapers
are still important political information sources for respondents regardless of gratifica-
tions categories.

The magnitude of gratifications demonstrated a' clear rank order among these
four information sources; however, it is also important to examine the relationships
among these media and their potential interchangeability as need satisfiers (Katz, Gure-
vitch, & Haas, 1973). Consequently, factor analysis is used to explore relationships
among these four‘media by incorporating a total of 44 items (11 items for each medium)
into the analysis.

The appropriateness of running a factor analysis on the data was pre-tested by
measuring sampling adequacy (KMO statistics). KMO statistics was calculated for vari-

ables in the data. High values indicate that a factor analysis is an appropriate method to

75

Table 10: Means Values and Standard Deviations of 11 Gratifications Items by

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Each Media
. . Candidate News News-
Gratlflcatlons Items Web Web paper TV

To see how the candidates stand . 2.43 2.68 3.51 3.73

on the issues, (1.39) (1.26) (1.03) (1.06)

9 To help me make up my mind on 2.09 2.51 3.45 3.62

5 how to vote in the election, (1.23) (1 .29) (1.10) (1.18)

‘3— To see what the candidates would 2.42 2.49 3.40 3.58

8 do ifelected. (1.45) (1.28) (1.11) (1.17)

. . . 2.20 2.41 3.41 3.75

To judge what candidates are Ilke, (1.34) (1.20) (1.06) (1.14)

:0 To remind me of my candidate’s 2.36 2.53 3.41 3.55

g, strong points, (1.41) (1.29) (1.11) (1.21)
8'

§ To get information which agrees 2.36 2.50 3.28 3.38

with my positions, (1.43) (1 .27) (1.13) (1.24)

. . . 2.00 2.45 3.31 3.38

c To use what I learn in politics, (1.18) (1.31) (1.21) (1.25)

‘5 To give me something to talk 2.00 2.36 3.53 3.80

about with other people, . (1.27) (1.36) (1.26) (1.24)

. . . . 2.18 2.51 2.73 2.87

_ Because it is interactive, (1.37) (1. 40) (1.29) (1.37)
3

§ Because it gives me the control 2.50 2.86 3.16 2.98

$21. over what and when I want to use, (1.49) (1.44) (1.23) (1 .29)
'5'

To participate in the Presidential 2.16 2.45 3.36 * 3.48

Election campaign, (1.34) (1.34) (1.23) (1 .27)

 

Note: Each gratifications item is followed by “I use from the following sources.”

76

use; the closer to l, the more appropriate it is. Specifically, a KMO statistic more
than .80 is appropriate; one less than .50 is unacceptable (Hutcheson & Sofroniou, 1999).
All KMO statistics for individual variables were over .80: candidate Web sites = .956,
news Web sites = .960, newspapers = .931, television news programs = .933.

A principal component analysis with varimax rotation identified seven factors
which have more than 1 eigenvalue. The eigenvalue shows the amount of variance ex-
plained by each principal component with the sum of the eigenvalues equaling the num-
ber of variables (i.e., 44). These seven factors accounted for 73 percent of the variance.
Interestingly, the first four factors are perfectly media specific (separated by each me-
dium): candidate Web sites, news Web sites, television news programs and newspapers.
In other words, each of these four factors consists of items purely from a corresponding
medium. Thus, the first four factors can be named Candidate Web Factor, News Web
Factor, Television Factor, and Newspaper Factor, respectively. This finding further
confirmed the media specific nature of political information gratifications: respondents
regarded each gratifications as similar within each medium. It also indicates that re-
spondents distinctly recognized each medium for political purposes. Contrary to the
previous findings, gratifications of political information may be very media specific.

Now, assuming that gratifications of political information is media specific, the
first four factors are used for the further analysis. Each of the four factors is comprised
of significant numbers of items. Specifically, both Candidate Web Factor and News
Web Factor are comprised of all 11 items; Television Factor and Newspaper Factor are
made up of nine and eight items, respectively. On the other hand, the other factors (5th
to 7th) are comprised of only three or fewer items eventhough they have eigenvalues of

more than one. Hence, these factors were dropped from the analysis. Table 11 shows

77

Table 11: Factor Analysis of Gratifications Items for Presidential Election
Information Sources

 

FACTOR 1: CANDIDATE WEB FACTOR

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Media Gratifications Items Factor
Loadings

To see how the candidates stand on the issues, .78
To help me make up my mind on how to vote in the 80
election, '
To see what the candidates would do if elected, .80

i1?

‘3: To judge what candidates are like, .82

D.

1:)

E To remind me of my candidate’s strong points, .82

to

a .

g2 To get information which agrees with my positions, .80

to

O . . .

§ To use what I learn in polltlcs, .76

a;

§ To give me something to talk about with other 72

g ' people, '

(0
Because it is interactive, .78
Because it gives me the control over what and 78
when I want to use, '
To participate in the Presidential Election campaign, .79
Percent of variance explained 32.48%
Eignevalue 14.29

. 2.24
Candidate Web Index Mean (SD)
(1.14)

Candidate Web Index Reliability (0) .96

 

 

78

Table 11: (Continued)

 

FACTOR 2: NEWS WEB FACTOR

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. . . Factor
Med la Gratificatlons Items L o a di n g 3
To see how the candidates stand on the issues, .80
To help me make up my mind on how to vote in the 82
election, '
To see what the candidates would do if elected, .81
2
§ To judge what candidates are like, .83
i To remind me of my candidate’s strong points, .83
1.0.
6' To get information which agrees with my positions, .79
G)
3%; Toause what I learn in politics, .69
3. To give me something to talk about with other 77
a people. °
Because it is interactive, .73
Because it gives me the control over what and
.71
when I want to use,
To participate in the Presidential Election campaign, .71
Percent of variance explained 20.22%
Eignevalue 8.90
News-Web Index Mean (SD) 2'51
(1.12)
News Web Index Reliability (0) .96

 

 

79

Table 11: (Continued)

 

FACTOR 3: TELEVISION NEWS FACTOR

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. . . Factor
Media Gratificatlons Items Loading 6
To see how the candidates stand on the issues, .80
To help me make up my mind on how to vote in the
. .79
. election,
To see what the candidates would do if elected, .80
.4
$- To judge what candidates are like, .76
(If
3 To remind me of my candidate’s strong points, .81
z
(D
3 To get information which agrees with my positions, .71
G)
g; To use what I Ieam in politics, .68
g. To give me something to talk about with other 55
a people, '
In
Because it is interactive, -
Because it gives me the control over what and _
when I want to use,
To participate in the Presidential Election campaign, .67
Percent of variance explained 5.86%
Eignevalue 2.58
. . 3.59
Televrsmn News Index Mean (SD) ( 92)
Television News Index Reliability (0) .91

 

 

80

. Table .11: (Continued)

 

FACTOR 4: NEWSPAPER FACTOR

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

. . . Factor
Media _ . Gratifications Items L oa dings
To see how the candidates stand on the issues, .73
To help me make up my mind on how to vote in the
_ , .70
election,
To see what the candidates woulddo if elected, .75
g To judge what candidates are like, .76
5 . ,
B To remind me of my candidate’s strong points, .80
'0
c ,
9 To get information which agrees with my positions, .72
a) - -
5’ To use what I learn in politics, .60
‘3. To give me something to talk about with other _
0’ people,
Because it is interactive, -
Because it gives me the control over what and _
when I want to use,
To participate in the Presidential Election campaign, .59
Percent of variance explained 4.76%
Eignevalue 2.09
3.39
Newspaper Index Mean (SD)
(.87) .
, Newspaper Index Reliability (0) .91

 

 

81

components and factor loadings for Candidate Web Factor, News Web Factor, Television
Factor, and Newspaper Factor.

Item responses for each. factor were first aggregated and then averaged to create
scale scores (indexes). Cronbach’s index of internal consistency (a) for each factor was
also calculated to confirm the adequacy of creating these indexes (see Table 11). The

first factor (Candidate Web Factor) was made into and named Candidate Index (M = 2.24,

SD = 1.14, a. .96); the second factor (News Web Factor) News Web Index (M = 2.51,
SD = 1.12, a = .96); the third factor (Television Factor) Television Index (M = 3.59, SD

= .92, a = .91); the fourth factor (Newspaper Factor) Newspaper Index (M = 3.39, SD

.87,a = .91). Values in all indexes range from 1 to 5.

Paired-sample t-test was used to determine the rank order of these indexes. All
the differences in these indexes are significant (difference between Television Index and
Newspaper Index: t = 5.35, df= 305, p < .001; Newspaper Index and News Web Index: t
= 12.75, df= 303,'p < .001; News Web Index and Candidate Index: t = 4.67, df= 303, p
< .001), which indicates that respondents’ gratifications for political information sources
are in the following rank order: television news, newspapers, news Web sites and candi-
date Web site. Television news was regarded as the most gratifying information source
for the presidential election campaign. while candidate Web sites were regarded as the
least.

Research questions asked whether the online media are alternative to tra-
ditional media. The findings in uses and gratifications demonstrated that the online
media did not become functional alternatives to the old media yet because respondents
did not regard the online media as gratifying as the traditional media. In addition, in-

terchangeability among these four information sources was not observed: the gratifica-

82

tions of political information is media specific. Taking these findings into consideration,
it is concluded that it takes a while for the online media to become an alternative to the
old media as a presidential information source.

Research Question 6 concerned whether respondents’ gratifications for the
online media are related to their political use of the online media (candidate Web
sites and news Web sites). Table 12 shows a correlation matrix between each index
and the respondents’ consumption of each information source. Each index is correlated
with a corresponding medium use, except for a couple of overlaps, which means that
these indexes are reliable predictors for corresponding media use.

These gratifications indexes may not be the only predictors for respondents’
media use. Demographic variables such as gender may affect the respondents’ informa-
tion use. Therefore, multiple regression was also employed to assess Research Question
6 by including gender, computer ownership, partisanship, party preference, online hour,
- and the gratifications indexes as independent variables. Gender (0 = female and 1 =
male), computer ownership (0 = do not own and 1 = own), partisanship (0 = non-partisan
and 1 = partisan), and party preference (0 = support a major party candidate and 1 =
support a third party candidate) were dummy-coded. The multiple regression can ex-
plore the multivariate relationships and the degree of impact of these independent vari-
ables on the use of each of the four information sources.

As for candidate Web sites, a preliminary analysis identified that Candidate In-
dex was the only significant predictor for the use of candidate Web sites. The Index
represents respondents’ gratifications level for the use of candidate Web sites. All of
the other variables were not significant predictors so that these insignificant variables

were dropped out of the regression analysis. The next analysis confirmed that logging

83

Table 12: Correlation Matrix between Gratifications Factors (Indexes) and Media

 

 

 

 

 

Use

FACTOR 1 FACTOR 2 FACTOR 3 FACTOR 4
Media Use Candidate News Web Television * Newspaper

Web Index Index Index Index
Candidate ” .1.
Web Sites .156 . .190 -.074 .011
News Web .085 .160“ -.098 .020
Sltes
Television

-.028 .008 239*“ .133*

News
Newspapers -.014 .022 .031 253*"

 

 

Note: *p < .05, **p < .01, p <

.001

84

 

on to candidate Web sites was predicted significantly by the Candidate Index (13 = .126, p
< .01) as is shown in Table 13. i i l l A

As was done for candidate Web site use, a preliminary analysis was conducted
to screen out insignificant variables from the analysis of news Web site use. It found
that online hour and News Web Index are possible candidates for the further analysis.
Multiple regression analysis was then conducted by including only these two variables.
The results are presented in Table 13. Online time spending was a significant predictor
for the use of news Web sites (B = .247, p < .001), while News Web Index was margin-
ally significant (B = .100, p = .08). Overall, the findings by multiple regression further
confirmed the reliability of gratifications indexes as predictors for respondents’ online
media use patterns.

Research Question 7 explored the difference in terms of Internet gratifica-
tions between’major party supporters and third party supporters. This study re-
ported ab0ve that third party supporters/voters were more likely to log on to candidate
Web sites than major party supporters/voters. Therefore, third party supporters/voters
are expected to appreciate the Internet over the traditional media more than major party
supporters. All of the four indexes——Candidate, News, Television, and Newspa-
per—were used to test the difference between these two groups.

Table 14 shows the mean values of these gratification indexes by major party
supporters/voters and third party supporters/voters. The difference of each gratifica-
tions index between the major party supporters and third party supporters was assessed
by t-test. All t-values were below significant level. Contrary to the expectation, there
was no significant difference in terms of gratifications between major party support-

ers/voters and third party supporters/voters.

85

Table 13: Regression AnalysisrResults Indicating Impact of Gratifications on
Candidate and News Web Site Use

 

 

 

Candidate Web Site Use
R R2 6 p
.156 .024 .006
Candidate Web Index (Factor 1) .126 .006

 

 

F (1, 304) = 7.58, p < .01

 

 

News Web Site Use
.288 .083 .000
Online Hour (hours) .247 .000
News Web Index (Factor 2) .100 .080

 

 

F (2, 334) = 13.69, p < .001

 

86

Table 14: Index Means and Major Party Candidate SupportersNoters v. Third
Party Candidate SupportersNoters

 

 

 

 

 

Major Party Third Party Major Party Third Party
Supporters Supporters Voters Voters
M (SD) M (SD) M (SD) M (SD)

Candidate 2.29 (1.14) 2.02 (1.16) 2.30 (1 .17) 2.40 (1.26)
'"dex t= .86, df= 262, ns t= .36, df= 210, ns
News Web 2.54 (1 .11) 2.61 (1.10) 2.57 (1.12) 2.76 (1.10)
"we" t= .23, df= 264, ns t= .77, df= 211, ns
Television 3.66 (.90) 3.47 (.98) 3.61 (.92) 3.34 (.89)
mm” t = .76, df= 262, ns t= .1 .28, df= 209, ns
Newspaper 3.41 (.87) 3.58 (.63) 3.43 (.84) 3.41 (.94)
Index

 

t=.71,df=262,ns

 

t=.11,df=208,ns

 

87

Summary of Results

H1: A person logs on to one Is preferred candidate 3' campaign Web site more often than

the other candidate sites.

Findings: The tendency of selective exposure to preferred candidate Web sites was gen-
erally observed in respondent’s Web use patterns. H1 was generally sup-
ported.

H2: The stronger the endorsement for a candidate a person has, the more likely and fie-

quently a person logs on to the candidate Web site.

Findings: Strong supporters were more frequently and likely to log on to their preferred
candidate Web sites than weak and middle supporters. Respondents logged
on to their preferred candidate Web sites according to not only whether or not
they endorse the candidates but also how much they endorse the candidates.
H2 was supported.

H3 .' The more a person logs on to the preferred candidate Web site, the stronger one is

endorsement for the candidate will become.

Findings: The analysis could not confirm that the exposure to preferred candidate Web
sites strengthened respondents’a endorsement for the candidates at Time 2.
Although it observed a weak indication of the reinforcement effect among
Nader voters, causal inferences cannot be drawn from the finding. Hypothesis
3 was not supported.

RQI .' Is there any relationship between logging on to candidate Web sites (or news Web

sites) and reported voting?

Findings: Users of candidate Web sites and news Web sites were more likely to vote than

non-users. This tendency held after controlling for other media consumption

88

 

and demographic variables. The result indicates that exposure to online in-

. formation sources about the presidential election motivates respondents to vote,
and the effect of candidate Web sites on voting is stronger than that of news
Web sites.

RQ2: Are third party candidate supporters more likely than major party candidate sup-

porters to log on to candidate Web sites and news Web sites?

Findings: Third party supporters/voters were more likely to use candidate Web sites than
major party supporters/voters. In terms of news Web site use, on the other
hand, there is no difl’erence at least statistically between third party endorsers
and major party endorsers. The results suggest that a difference exists be-
tween third party endorses’ use of candidate Web sites and their use of news
Web sites.

RQ3: What gratifications do individuals seek from online campaign Web sites?

Findings: Respondents did not seek any specific gratifications from candidate Web sites
and news Web sites, rather they regarded each gratifications item as similar
within each medium.

RQ4: Are there any diflerences in gratifications people seek fiom Web sites and other in-

formation sources?

Findings: Respondents clearly distinguished the online media from the old media.
Their gratifications for political information sources are in the following rank
order: television news, newspapers, news Web sites and candidate Web site.

RQ5: Is the Internet an alternative to traditional media?

Findings: The online media did not become functional alternatives to the old media yet

because (1) respondents did not regard the online media as gratifying as the old

89

media and (2) interchangeability among these four information: sources was not
observed.
' RQ6: Are the uses and gratifications for Web use a predictor for online use?

Findings: Gratifications indexes for the four information media are correlated
with corresponding media, which means that these indexes are reliable predictors not
only for the online media use but also the old media use.

RQ7: Is there any difierence in Internet uses and gratifications between major party sup-
porters and third party supporters?

Findings: There was no significant difference in terms of gratifications between

major party supporters/voters and third party supporters/voters.

90

CHAPTER V

DISCUSSION

Possible Impact on Voting Behavior

This study presented new and different empirical evidence that candidate Web
site use and news Web site use, as opposed to traditional media use, are significant pre-
dictors for the higher probability of voting, as Charts 7 and 8 clearly demonstrated. Past
research found that the use of the traditional media such as viewing television and read-
ing newspapers is an indicator of higher voting turnout. However, the effects of televi-
sion news and newspapers on voting were partialled out after controlling for the use of
candidate Web sites and news Web sites. This result has an important implication: Web
site use for political purposes may have a tremendous impact on voting and the influence
of the traditional media on voting may be fading away in the lntemet age.

The results in uses and gratifications seem to contradict this possible effect of
the Web sites. The findings of this study demonstrated that respondents’ gratifications
for the online information sources were ranked after newspapers and television news
programs. While television news programs and newspapers generally gratified respon-
dents much more than candidate Web sites and news Web sites did, these traditional me-
dia were not found to have effects on voting turnout in this study. This complicated re-
lationship between the media, gratifications, and voting likelihood raises a question: Why
is it that, despite that these online media were predictors for voting, candidate Web sites
and news Web sites did not gratify respondents as much as the traditional media did?

This study did not verify cause and effect evidence that logging on to the Web

sites actually led respondents to the polls. It is entirely possible that respondents who

91

 

have already decided to vote are more likely to log on to the Web sites than those less
determined, i.e., the pro-determination proposition. If the pre-determination proposition
is right, the decision to vote should nullify the online media’s effects on voting turnout in
a multivariate analysis; however, the proposition could not invalidate the possibility that
the Web site use has an impact on higher voting. Another multiple logit regression
analysis was conducted by including the “preedetermination variable” (Time 1 decision
to vote for a candidate: coded 0 for No, 1 for Yes) as well as these online media use as
independent variables. The result demonstrated that candidate Web sites and news Web
sites still have significant impacts on the higher probability of voting after controlling for
“pm-determination.”

A plausible mechanism of the Web site impact on voting resides in Web sites’
uniqueness: the Web sites can (1) provide consonant content for visitors and (2) give
them a sense of community. First, you can selectively choose what you want to see
and/or read through the lntemet. Thus, it is quite possible that you can obtain content
online which is consistent with your ideas. A candidate Web site is a good example.
The Web site provides only positive information about the candidate. All the informa-
tion on the Web site is designed to affirm the strength and superiority of the candidate
over other candidates. Second, a candidate Web site, or any Web site designed for spe-
cific types of people can give visitors a sense of community (for example, Rheingold,
1999). Sense of community can be defined as a feeling of being part of the community
and a feeling of sharing common interests with community members. However, the
traditional media such as television, radio and newspapers are not. structured to bring
about a sense of community because of one-way, top-down communication (Rucinski,

1991). People can virtually visit a community on the Web or even become a member of

92

the cyber community not only by communicating and interacting with each other but also
by tailored content of Web sites for specific visitors.
C yber Motivation Hypothesis

Synthesizing these two elements, I would like to propose the cyber motivation
hypothesis which posits that (1) selective exposure to consonant messages on a Web site
enhances confidence, or self efficacy, (2) the Web site offers a sense of community which
enhances self-efficacy, and (3) the obtained and enhanced self-efficacy eventually affects
voting behavior, i.e., motivates people to vote. The mechanism of how this model
works is elaborated as follows.

After people selectively expose themselves to ideas and messages consistent
with their own, they feel reaffirmed and then their commitment is strengthened. The
confirmation eventually enhances their confidence about what they will actually do.
This confidence in behavior is known as self-efficacy, which refers to the confidence
people have in their ability to do what they want (Bandura, 1986), and it is a pivotal fac-
tor of behavior change and actiOn (Maibach & Cotton, 1995). In addition, a community
can provide efficacy information which can have a strong impact on the community
members’ efficacy beliefs. This is called “vicarious efiicacy information” (Maibach &
Cotton, 1995, p. 48). Thus, visitors to a specific Web site could enhance self-efficacy
by the “double dose” of efficacy boosters: (1) selective exposure to consonant messages
and (2) feeling a sense of community. Subsequently, the outcome of enhanced
self-efficacy motivates them to take action. The cyber motivation hypothesis refei's to
this series of effects from cognition, to attitude and finally to behavior (see Figure 1).

Let’s apply this hypothesis to the case of candidate Web sites. First, voters se-

lectively expose themselves to a preferred candidate Web site. Then, they will reaffirrn

93

 

£56860

 

 

 

 

 

 

 

 

 

 

co emcow
mc=o> 38.5.2.3 . . 3w nc>> 93896
Long: 605350 23630 82620.4).
commode—2
Encomcoo

 

awe—=82... c2628.). Logo ._oeo_2 Scum acumoom .mogw no>> new oaamoaxm o>_6c_ow “F 359“.

94

their belief or choice for the candidate by supporting messages and content on the Web
site. At the same time, they will feel a sense of community there because the site is
structured for a specific purpose and people. , Consequently, they will become more
self-efficacious by the double dose of these eflicacy boosters and motivated to take a ac-
tion, thus voting for the candidate.

The cyber motivation hypothesis does not try to invalidate the antithetical
proposition, namely, the pre-determination proposition. It is rational that both mecha-
nisms influence each other: those who have already determined to vote may be further
motivated by boosting self-efficacy. Without any empirical evidence, however, it is safe
to say that these two mechanisms, the pre-determination and the cyber motivation, influ-

ence and interact with each other to raise the possibility of voting.

Selective Exposure to Candidate Web Sites

The findings regarding selective exposure generally confirmed respondents’
tendency toward preferred candidate Web sites. In addition, this tendency becomes
stronger as the strength of respondents’ support for the candidates increases. Thus, it
can be concluded that the strength degree of pre-disposition determines the degree of the
selective exposure to candidate Web sites. This is an important finding, but not a sur-
prising one, which provides the literature with a piece of empirical evidence for the se-
lective exposure theory in the lntemet age. The findings also confirmed a characteristic
of the lntemet: information on Web sites is the “pull” variety, i.e., the selection of online
information is in the users’ hands.

Even though the tendency toward selective exposure was confirmed, the results

suggest that selective exposure to a preferred candidate Web site did not boost respon-

95

 

dents’ endorsement for the candidate. Of course, this does not necessarily mean that
selective exposure cannot boost the degreeof support, but it means that this study could
not find any evidence which indicates a booster effect of Web site exposure on endorse-
ment degree. In fact, the findings on voting behavior and selective exposure infer that
the use of candidate Web sites has a boosting effect on endorsement.

This study found a possible correlation between the support increase/decrease
from Time 1 to Time 2 and the frequency of candidate Web site use among Nader sup-
porters (r = .53). The association was not statistically significant due to the small sam-
ple size. However, considering that (1) information about third party candidates is
scarce in the main stream media and then (2) third party supporters are more likely to use
candidate Web sites than major party supporters are, there is a possibility that third party
supporters increased their endorsement at least partly by the exposure to candidate Web
sites. Thus, the possibility of the boosting effect by selective exposure should not be
eliminated because this-sample size was not large enough.

In the selective exposure section, there is a mixed finding which should be in-
terpreted in the context of the presidential election 2000. While selective exposure was
generally supported by the results, Nader supporters were somewhat ambivalent toward
the use of the Nader and Gore Web sites. In terms of frequency, Nader supporters vis-
ited the Nader Web site (1.47 times) more often than the other sites (.94 times for Gore’s
and .59 times for Bush’s). However, in terms of probability, they were almost equally
likely to log on to the Nader site and the Gore site (47.1 % and 41.2 %, respectively)
while less likely to log on to the Bush site (29.4 %). This seems to be consistent with
the often-reported news story that Nader supporters and Gore supporters, not Bush sup-

porters, somewhat overlapped. Nader supporters may not have been able to make a de-

96

cision about which candidate, Gore or Nader, they would vote for during the election
campaign. ’ Thus, it could be said that the close probabilities are attributable to this in-
decisiveness.

The descriptive statistics in the selective exposure section showed that Bush
supporters used Web sites less than the other candidate supporters did. A plausible in-
terpretation is that while Gore and Nader supporters tend to be Internet sawy, Bush sup-
porters are less likely to use the lntemet. A statistical analysis was conducted to test
whether or not Gore and Nader supporters were more likely to log on to candidate Web
sites than Bush supporters. In terms of probability, 65% of Nader supporters and 36%
of Gore supporters used any candidate Web sites while 27% of Bush supporters did so
( x 2 = 10.29, p < .01). In terms of frequency, Nader supporters used any candidate Web
sites 3 times throughout the campaign, while Gore supporters did so 2.03 times and Bush
supporters 1.50 times. While the difference in frequency is substantive, it was not sta-
tistically significant (F [2, 219] = 1.40, ns.) because the standard deviations of logging on
frequency are huge. The results are not outright, however, party identification might be

a factor to determine the use of candidate Web sites among respondents.

Third Party Supporters’ Web Use

This study found that third party supporters, Nader supporters in this study, were
more likely to log on to candidate Web sites than major party supporters. Third party
supporters’ greater use of candidate Web sites seems to be attributable to a feature of the
lntemet: people can obtain what they want to see and hear on the lntemet. On the other
hand, there was no significant difference in terms of news Web site use between third

party supporters and major party supporters. This finding raises a question: Why is it

97

that the difference between third party and major party supporters was observed only in
the use of candidate Web sites?

One plausible interpretation is that information content on news Web sites is
nearly identical with that in the traditional media because almost all news Web sites are
run by and/or provided with news content by their “parents,” maj or news organizations.
For this reason, it is probably difficult for third party supporters to obtain information
about third party candidates through online news sites, but it is easy through the candi-
date Web site. Thus, it can be said that Nader supporters logged on to candidate Web
site out of necessity.

Another interpretation may be that third party supporters as a whole are more
likely to be “Internet geeks,” or in politically correct language, lntemet savvy. However,
" according to the data, this interpretation does not seem to be legitimate. The differences
in terms of overall consumption on the lntemet between major party supporters/voters
and third party supporters/voters were quite similar. At Time 1, third party supporters
spent an average of 9.81 hours (SD = 8.80) online a week while major party supporters
10.12 hours (SD = 9.22). At Time 2, third party voters went online an average of 11.05
hours (SD = 11.50) while major party voters 10.22 hours (SD = 8.89). These differ-
ences are not statistically significant (the former: t = .13, us; the latter: I = .40, ns).
Thus, as stated above, third party supporters/voters seem to need to log on specifically to

candidate Web sites.

Media Specific Uses and Gratifications
Contrary to the findings in previous studies, respondents’ uses and gratifications

about political information sources are very media specific. The magnitude of the grati-

98

fications for each medium demonstrated the rank order among the media in terms of po-
litical information sources. Television news programs gratified respondents most, fol-
lowed by newspapers. On the other hand, online information sources such as candidate
Web sites and news Web sites were least appreciated. This finding casts doubt on what
has been taken for granted in uses and gratifications research, i.e., non-media specific
nature of uses and gratifications in political information sources. Perhaps the new me-
dia environment with the advent of the Internet is responsible for the media specific na-
ture of uses and gratifications in a way that each medium found a “niche” for itself.

Past studies that found non-media specific nature of political information grati-
fications were conducted when the media meant newspaper, radio and television, i.e.,
before the inception of the lntemet. Media variety in the past was not as varied as it is
now. Today, the lntemet has established a foothold in the media environment and is
becoming as an important medium as the traditional media. For example, in terms of
time spent on each medium, respondents went online an average of 10 hours a week,
while they spent only a fifth or less of the time on the other media such as television and
newspaper. The Internet was the most extensively used medium by respondents, al-
though this does not necessarily mean that they used candidate Web sites and news Web
sites as much as they used the lntemet as a whole. The clear rank order, then, may be
attributed to distinct characteristics of the old media and the new media. The differ-
ences are evident because (1) the lntemet is a “pull” medium while the old media are
“push” media and (2) the Internet is the latest addition to the media environment so that
its newness can draw more attention. Thus, for these reasons, respondents may have

been able to distinguish and rank-order candidate Web sites, news Web sites, television

news programs and newspapers.

99

 

 

The results also indicated that the old media are still important political informa-
tion sources for respondents regardless of gratifications categories. However, consider-
ing that candidate Web sites and news Web sites appeared to have had an impact on vot-
ing, it would be wrong to conclude that the online media are trivial merely because they
were ranked as distant third and fourth places.

The online media’s lower gratifications could be attributed to the fact that only a
few respondents logged on to the Web sites. The overall gratification levels for the
online media (Candidate Index and News Web Index) were low because these indexes
were the mean values of all respondents. Specifically, of all respondents at Time 1,
only 41 of them actually logged on to a candidate Web site(s). Not surprisingly, the
candidate Web site users’ gratifications level is greater than that of non-users. Their
gratifications for candidate Web sites (Candidate Index) is 3.0 compared with 2.12 of
those who did not use, and this difference is statistically significant (t = 4.79, p < 001).
As presidential candidates will probably utilize the Internet to construct more attractive
Web sites in the 2004 election and as more people use candidate Web sites, the level of
gratifications for candidate Web sites is expected to go up.

Overall lower gratifications levels for the online media were at least partly ex-
plained by the small number of users. However, there remains a question which is not
explained for this reason: even though candidate Web sites and news Web sites have a
unique feature of interactivity, these lntemet media were rated lower even in the interac-
tive items than television news programs and newspapers which do not have that func-
tion. For example, candidate - Web sites have such interactive features as “real time
question and answer” and “email or contact us.” So it'had been expected that candidate

Web sites would have been rated highest in the interactive gratifications items or at least

100

 

as high as the other media. Contrary to the expectation, it was found that candidate
Web sites were perceived as the least gratifying political information source even in the
interactive items. Aside from the small number of users, the result suggested that these
interactive functions failed to make respondents feel like they really interacted with the
candidate.

The reason that candidate Web sites were least gratifying could also be ex-
plained by skepticism about the sites. Respondents may be doubtful about or even dis-
trust that the candidate would read and respond himself. Thus, while the interactive
function can provide them with quick response and feedback online, people may feel de-
ceived or suspicious that the staff members respondrather than the candidate (Park &
Choi, 2001). In- this way, skepticism about the reality of interactivity may have nullified
the Intemet’s unique interactive function. Another interpretation is that respondents did
not regard these interactive features as interactive afier all. Generally speaking, such
functions as “real time question and answer,” “chat room” and “online contribution” are
technically labeled as interactive features; however, there may be a gap in terms of the
definitions of “interactive features” between respondents’ conception of interactive fea-
tures and general labeling of interactive features. Thus, it can be said that the lntemet

still has a long way to go for its interactive features to be actually regarded as interactive.

101

> CHAPTER VI

CONCLUSION

Contributions and Appropriateness

New technologies have continuously brought new channels of political commu-
nication to us. Originally in US. political campaigns, the media meant print materials,
such as newspapers and flyers. Later, magazines, radio news and advertising joined the
campaign media. Television played a significant role in the modern campaign in the
second half of the 20th century. Now the latest technology in political communication
is the lntemet, which has made interactive online campaigning possible. As television
has changed the political process in the modern election campaign, the lntemet is ex-
pected to change the political process and play as significant or even a more significant
role in election campaigns than television (for example, Morris, 1999).

Interestingly, little was actually known about how and why voters use Web sites
to retrieve political information and to interact with candidates. This dissertation was
inspired by this lack of knowledge. I started this dissertation to fill the void of under-
standing by conducting a panel survey which could capture a longitudinal glimpse of
voters and verify any possible cause and effect relationships. The two-wave panel sur-
vey was conducted at universities in Michigan, Missouri and Texas between September
and November 2000. A total of 325 college students participated both in Time 1 and
Time 2. Ideally speaking, the number of respondents may not be enough for the study
of a presidential election because this topic is the central focus on American political be-
havior (Cavanaugh, 1995). However; considering most social science research includ-

ing political communication has been done with one-time measurement, the panel-survey

102

 

 

probably compensated for the shortage.

The dissertation has documented five major findings: (1) a potential. for Web
sites to enhance the probability of voting turnout, (2) respondents’ tendency of selective
exposure toward preferred candidate Web sites, (3) third party supporters’ dependency on
candidate Web sites, (4) the media-specific nature of political uses and gratifications and
(5) respondents’ tendency to be more gratified by the traditional media than by the online
media.

The first major finding provided empirical evidence contrary to what had been
found in the previous studies, i.e., the online media were predictors for voting turnout
while the traditional media were not. This finding allowed me to construct the cyber
motivation hypothesis, which posits that the Web’s efficacy boosting roles eventually
enhance the likelihood of voting. The second major finding also includes evidence that
the degree of preference for a candidate corresponded to the degree of selective exposure.
While this study could not confirm the cause and effect relationship that selective expo-
sure actually boosts the degree of endorsement, this evidence is another major finding.

The third should be regarded as good news for an ideal form of democracy in
which everybody has an equal opportunity for political information regardless of political
identifications. Third party supporters’ tendency toward Web sites could be interpreted
as a way that they can obtain precious political information about their candidates, which
is otherwise difficult to get. From a different viewpoint, the lntemet can be seen as a
provider of the equal information means for third party supporters because, on the Inter-
net, every candidate can have the same opportunity to present information for voters

everywhere.
The fourth and fifth findings about uses and gratifications provided important

103

empirical evidence in the lntemet age. The former finding demonstrated that respon-
dents distinguished each medium as a political information source and also indicated that
each medium in the Internet age has a niche for itself. The latter finding represented the
low overall status of Web sites, especially candidate Web sites, in the presidential elec-
tion 2000. The status of Web sites as a political information source in the presidential
election 2004 depends both on how much candidate campaign camps exploit the lntemet
and how much voters utilize the Web for political information sources. For a compari-
son, this study provided an essential benchmark with which the gratifications. status for
the Web sites can be measured in future research.

The above findings as a whole shed light on the online media’s roles in the
lntemet age presidential election. Therefore, these findings augmented the theoretical
scope on the Web sites and helped understand the mechanism of selective exposure and
its possible consequence on voting turnout. They also can help improve presidential
candidate Web sites by providing new empirical evidence. Together, these are the major
contributions to the literature.

This dissertation employed selective exposure and uses and gratifications para-
digms because the combination was judged most appropriate for the examination of cy-
ber campaigning from the perspective of voters. One of the reasons for the choice is
that uses and gratifications incorporates audience motivation and behavior for a specific
media use. That information on the Web is “pull” is the reason for the choice of selec-
tive exposure and should have important implications in an election campaign.

The other reason for the choice is that it is important to study the differences
between online campaigning and traditional campaigning. Uses and gratifications could

distinguish voters’ choice of the Internet from that of traditional channels because studies

104

using the theory can include both the lntemet and traditional channels as variables. As
presented, the results clearly distinguished the difference between candidate Web sites,
news Web sites, television news programs and newspapers. Thus, it can be said that the

approach this dissertation took was successful in producing expected and necessary data.

Implications on Presidential Election Campaigning

This dissertation presented an important implication for future presidential elec-
tions: Web site use may have a tremendous impact on voting turnout. It also demon-
strated the characteristics of candidate Web site visitors: the visitors are more likely to be
supporters. How can these aspects be utilized in a presidential election campaign and
why?

The most important roles of candidate Web sites is to get visitors to decide to
vote for the candidate (Reavy, 1999). The tendency toward selective exposure will pro-
vide presidential election campaign management with an important clue about candidate
Web site visitors—that they are most likely those who already endorse the candidate.
Thus, Web masters for presidential elections are able to customize the content of the Web
sites to target their audience. Targeting the audience is very important when making
Web sites because effective content leading to behavioral decision-making differs sig-
nificantly according to where the audience stands, in terms of what stage they are in (Di-
Clemente & Prochaska, 1985; Prochaska, DiClemente, & Norcross, 1992).

According to the stage theory, there are three ranked stages leading to action:
pre-contemplative, contemplative, and preparation (Holtgrave, Tinsley & Kay, 1995).

Pre-contemplative refers to the first stage, at which one does not yet recognize the need

to take an action. In the next stage, contemplative, one seriously thinks about the action.

105

I‘ 4". ta —

 

 

In preparation, the last stage prior to the action, one is prepared for the action. This
study’s findings suggest that those who visit candidate Web sites are more likely to have
considered voting for (contemplative stage) or have already decided to vote for a candi-
date (preparation stage). Since those in the preparation stage have already decided to
vote, Web masters may not need to target Web visitors in this stage. The logical target
audience, then, is visitors in the contemplative stage. Web masters can construct effec-
tive sites to advance the audience from the contemplative stage to the preparation stage
(see Holtgrave, Tinsley & Kay, 1995 for effective message/content construction).

Should Ralph Nader run forthe presidency in 2004, the Green Party campaign
camp should utilize the lntemet as much as possible. His supporters were found to log
on more to his campaign Web site than other candidates’ supporters, whereas they were
not necessarily heavy lntemet users. Considering that the news organizations seldom
covered Nader and that the Nader. Web site was one of a small number of information
sources for Nader supporters, the relative importance of the Web site outweighs that of
major party candidates.

It should be noted that a third party candidate and a major party candidate are all
equal in cyberspace, at least technically, with the same large potential audience. In fact,
contrary to what it should be, the Nader Web site was the least “information rich” one
compared to those of Bush and Gore as demonstrated in Chapter I. A third party carn-
paign camp cannot afford to lose supporters/voters who could be obtained through the
construction of a good Web site. Regardless of the Nader campaign camp’s lack of fi-
nancial resources to build a sophisticated Web site or indifference in the lntemet itself, it

is a logical step to provide sufficient information for his information-poor supporters.

106

 

Limitations and Recommendations

The findings of this dissertation, however, should be further confirmed by em-
pirical evidence from a variety of demographic groups in the coming election. First,
respondents participated in this study were all college students. Respondents provided
neither a sufficient representation of the US. citizen nor of the general population.
Even though this participant category has an advantage over other demographic groups
because they are all wired, the results obtained from the demographic group should be
understood with caution.

Although this study successfirlly administered the panel survey to examine the
long-term effects of Web sites, no such panel study is problem-flee. For example, this
panel study surveyed respondents twice, Time 1 in September and Time 2 in November.
Data which might have been obtained before Time 1 and in between Time 1 and Time 2
may provide more insightful and useful results leading to cause and effect evidence.
Accordingly, in such cases, these effects may be lost or misinterpreted.

Further, the number of respondents was another shortcoming, especially for the
analysis about third party supporters/voters. Since respondents identified as third party
supporters/voters consisted only a small portion of the total respondents, the shortcoming
consistently plagued the analyses. For example, this study found a possible correlation
between the support increase/decrease from Time 1 to Time 2 and the frequency of can-
didate Web site use among Nader supporters. While the correlation was substantive, it
was not statistically significant due to the small sample size. Thus, the boosting efl'ect
of selective exposure among third party supporters should be investigated again by future
research with an adequate number of subjects. .

Lastly, considering ever changing features of the lntemet and the media envi-

107

ronment as a whole, the findings in the present dissertation may not be applicable in the
next presidential election 2004. It is predicted that the online media will play more
significant roles in the election. How much more of a role will the online media play?
No one knows what the exact structure of the media environment will become or how the
online media will have evolved in 2004. The lntemet itself is transfiguring itself from a
simple interactive medium to 3D (three dimensional) and virtual reality-capable medium.
As the characteristics of the lntemet will transform, the way people use it will surely
change. The lntemet may become equivalent to the regular media or even ascend to the
throne which television now holds, as television took away the status in election cam-
paigns from newspapers in the past. It is necessary for researchers to constantly update

the roles of the lntemet in presidential election campaigns.

108

 

 

APPENDICES

 

109

Appendix A

Time 1 Survey: Information Use on the US Presidential Election 2000

This questionnaire asks you how you use information sources to make voting decisions in the
Presidential Election 2000. It takes a few minutes to complete this questionnaire. You will
be asked again to take a similar questionnaire for comparison purposes in November, after
the presidential election. This questionnaire asks you to write your name. however, your
response will be completely confidential. The participation of this survey is voluntary. You
indicate your voluntary agreement to participate by completing and returning this question-
naire.
* U.S. citizens only
1. Please print your name (For comparison purposes with the second questionnaire only.
We will replace your name with case 10 number. All completed questionnaires will be
destroyed after the end of this project).
First Last

 

2. Which candidate will you vote for if the election is held today? Circle one.
George w. Bush Al Gore Other (specify) Neither

3. Please mark the degree of support for the candidate you selected in the above question.
If you chose “Neither” in question 2, skip this question.
A little Support (circle one) Strong Support
1 2 3 4 5

4. Please indicate your partisanship regardless of your candidate preference and official
membership.
Democrat . Republican Other (specify ) Neither

5. Al Gore (www.algore2000.com), George W Bush (wwgeorgewbusheom) and other
candidates have their official campaign web sites. Estimate how often you have visited to
each candidate official web site. If you have not logged in to either candidate site, please
write “0" in both.

Bush site times Gore site times
Other site (specify ) times

6. In order to obtain information about the presidential election 2000. estimate the total num-
ber of times you have spent to visit news web sites such as the New York Times
(www.nyt.com) and CNN (www.cnn.com).

times

7. In the last 7 days, estimate how much time you spent watching or reading the following

. news sources for any pgmse.
Newspapers __ hours
News magazines __ hours

110

 

Television news programs hours
lntemet news sites hours
Radio news (such as NPR) hours

8. We want to find out why people use various political information sources for the presiden-
tial election. Here are some reasons that other peepie gave us for using particular infor-
mation sources. Please indicate how much you agree or disagree with the following
statements for each information source on a scale of 1 to 5, where 1 means strongly dis-
agree and 5 means strongly agree.

Example
Let’s take a statement “To see what kind of sports the candidates play, I use
information from the following sources,” as an example. If you strongly agge
with “To see what kind of sports the candidates play" by using candidates’ web
sites, please circle 5, which means strongly agree. if you diggree with “To see
what kind of sports the candidates play” by using news web sites, please circle
2, which means disagree.

 

 

“To see what kind of sports the candidates play, i use information from the fol-

 

 

lowing sources.” ,
Strongly Disagree (circle one) Strongly
Agree
Candidates’ campaign web sites 1 2 3 4 5
News web sites 1 2 3 4 5

 

8a. “To see how the candidates stand on the issues, I use information from the fol-
lowing sources.”
Strongly Disagree (circle one) Strongly Agree

Candidates’ campaign web sites 1 2 3 4 5
News web sites 1. 2 3 4 5
Newspaper 1 2 3 4 5
Television news 1 2 3 4 5

8b. “To help me make up my mind on how to vote in the election, 1 use information
from the following sources.”
Strongly Disagree (circle one) Strongly Agree

Candidates’ campaign web sites 1 2 3 4 5
News web sites 1 2 3 4 5
Newspaper 1 2 3 4 5
Television news 1 2 3 4 5

8c. “To see what the candidates would do if elected, I use information from the fol-
lowing sources.”

111

Strongly Disagree (circle one) Strongly Agree

Candidates' campaign web sites 1 2 3 4 5
News web sites 1 2 3 4 5
Newspaper 1 2 3 4 5
Television news 1 2 3 4 5

8d. “To judge what candidates are like, I use information from the following

 

sources.”
Strongly Disagree (circle one) Strongly Agree
Candidates’ campaign web sites 1 2 3 4 5
News web sites 1 2 3 4 5
Newspaper 1 2 3 4 5
Television news 1 2 3 4 5

 

8e. “To remind me of my candidate’s strong points, I use Information fromthe fol-
lowing sources.”

Strongly Disagree (circle one) Strongly Agree

Candidates’ campaign web sites 1 2 3 4 5
News web sites 1 2 3 4 5
Newspaper 1 2 3 4 5
Television news 1 2 3 4 5

8f. “To get information which agrees with my positions, I use Information from the
following sources.”
Strongly Disagree (circle one) Strongly Agree

Candidates’ campaign web sites 1 2 3 4 5
News web sites 1 2 3 4 5
Newspaper 1 2 3 4 5
Television news 1 2 3 4 5

89. “To use what I learn in politics, I use information from the following sources.”
Strongly Disagree (circle one) Shongly Agree

 

Candidates’ campaign web sites 1 2 3 4 5
News web sites 1 2 3 4 5
Newspaper 1 2 3 4 5
Television news 1 2 3 4 5

8h. “To give me something to talk about with other people, I use information from
the following sources.”
- Strongly Disagree (circle one) Strongly Agree

Candidates’ campaign web sites 1 2 3 4 5 ,
News web sites 1 2 3 4 5
Newspaper 1 2 3 4 5
Television news 1 2 3 4 5

8i. “Because it is interactive, I use information from the following sources.”

112

Strongly Disagree (circle one) Strongly Agree

Candidates' campaign web sites 1 2 3 4 5
News web sites ' ’ 1 - 2 3 4 5
Newspaper 1 2 3 4 5
Television news 1 2 3 4 5

8i. “Because it gives me the control over what and when I want to use, I use infor-
mation from the following sources.”
Strongly Disagree (circle one) Strongly Agree

Candidates’ campaign web sites 1 2 3 4 5
News web sites 1 2 3 4 5
Newspaper 1 2 3 4 5
Television news 1 2 3 4 5

8k. “To participate in the Presidential Election campaign, I use Information from the
following sources.” - .
Strongly Disagree (circle one) Strongly Agree

Candidates’ campaign web sites 1 2 3 4 5

News web sites 1 2 3 4 5

Newspaper 1 2 3 4 5

Television news - 1 2 3 4 5
9. Do you have your own computer with lntemet access? Yes No
10. How much time do you usually spend online per week? hours

11. How comfortable are you in using a computer? (Circle one)
Very Uncomfortable Very Comfortable
1 . 2 3 4 5

12. How comfortable are you in using the lntemet? (Circle one)

Very Uncomfortable Very Comfortable
1 2 3 4 5

13. Gender Male Female

14. Age years old

15. Level (circle one)
Freshman Sophomore Junior Senior Graduate

16. Major (please print)

 

Thank you very much

113

Appendix 3
Time 2 Survey: Information Use on the US Presidential Election 2000

This questionnaire asks you about voting decisions in the Presidential Election 2000. You
are asked to write your name, however, your response will be completely confidential. The
participation of this survey is voluntary. You indicate your voluntary agreement to participate
by completing and returning this questionnaire.
"' U.S. citizens only
1. Please print your name (For comparison purposes with the first questionnaire only. We
will replace your name with case ID number. All completed questionnaires will be de-
stroyed after the end of this project).

First Last

 

 

2. Which candidate did you vote for on November 7? Please circle one.
George w. Bush Al Gore Other (specify ) Did Not Vote

3. Please mark the degree of support for the candidate you selected in the above question.
If you chose “Did not vote” in ggestion 2, skip mis Question.

A little Support (circle one) Strong Support
1 2 3 4 5

nl for those wh “Di Not V ,” which candidate would you have voted for if you had
had a chance? And mark the degree of support for the candidate you selected here. I;

you actually voted, skip this Question.

 

George w. Bush Al Gore Other (specify ) Neither
A little Support (circle one) Strong Support
1 2 3 4 5

5. Al Gore (www.algore2000.com), George W Bush (www.georgewbush.com) and other
candidates have their official campaign web sites. Estimate how often you have visited to
each candidate official web site during the Presidential election campaign. If you have not
logged in to either candidate site, please write “0" in both.

Bush site times Gore site times Other site (specify ) times
5. In order to obtain information about the presidential election 2000, estimate the total num-
ber of times you have spent to visit news web sites such as the New York Times

(www.nyt.com) and CNN (www.cnn.com). times Thank You!

Please Make Sure You Have Written Your Name on This Questionnaire

114

 

 

Appendix C
Code Book
Time 1 Survey Variables
NOTE: Bold variable names stand for original vs; italic variable names stand for recoded vs.

id Identification numbers of respondents

100s Saint Louis University

200s University of Texas, San Antonio (201-231 =1)
300s MSU. ADV417

4003 MSU. ADV473

500s MSU. COM3

600s MSU. JRN

700s- MSU, ISSZZS (701-850=1)

panel Participation in the panel survey
0 = Time 1 and 2 only
1 = Time 1, 2 and 3

vote‘l

Which candidate will you vote for if the election is held today?

1 = Bush 2 = Gore 3 = Nader 4 = Other 5 = None
vote 1x ‘

1 = Nader O = Bush 8. Gore 9 = Missing (undecided 8 other third party)
vote1xx

1 = Third Party 0 = Bush 8. Gore 9 = Missing (undecided)

votel y

1 = Any candidate 0 = Undecided 9 = Missing

votebng

1 = Bush 2 = Gore 3 = Nader 9 = All others
supporti

Please mark the degree of support for the candidate you selected in the above question. If
you chose “Neither" in question 2, skip this question.
1 = a little support to 5 = strong support

sup1_123 Supporters are divided into three sub-groups, weak (1 82), middle (3), and
strong supporters (4 85). .

1 = 1 82 2 = 3 3 = 4 8 5

partisan

Please indicate your partisanship regardless of your candidate preference and official
membership.

1 = Democrat 2 = Republican 3 = Green 4 = Other 4 = None

partix Partisan is divided into two, categories:
1= partisan 0 = non-partisan

115

Iogbush1 Open-ended

Ioggore1 Open-ended
Iognader1 - Open-ended
logoth1 Open-ended

Al Gore (www.algore2000.com), George W Bush (www.9eorgewbushcom) and other
candidates have their official campaign web sites. Estimate how often you have visited to
each candidate official web site. If you have not logged in to either candidate site, please
write "0” in both.

Log1
The aggregation of Iogbush1 through logoth1

Log1bi
Based on Iog1, whether or not one has logged in to any candidate site.
1 = at least 1 time 0 = Zero

Iognews1 Open-ended

In order to obtain information about the presidential election 2000, estimate the total number
of times you have spent to visit news web sites such as the New York Times (www.nyt.com)
and CNN (www.cnn.com).

paperuse Open-ended
magazuse Open-ended
tvuse Open-ended
netuse — Open—ended
radlouse Open-ended

In the last 7 days, estimate how much time you spent watching or reading the following news
sources for any purpose.

8a. “To see how the candidates stand on the issues, I use information from the following
sources.”

a_cand 1 - 5 Likert scale
a_web 1 - 5 Likert scale
a _paper 1 - 5 Likert scale
a_tv 1 — 5 Likert scale

8b. “T 0 help me make up my mind on how to vote in the election, I use information from the
following sources.”

b_cand 1 - 5 Likert scale
b_web 1 - 5 Likert scale
b_paper 1 - 5 Likert scale
b_tv 1 — 5 Likert scale

8c. “To see what the candidates would do if elected, I use information from the following
sources.”

c_cand 1 '- 5 Likert scale
c_web 1 - 5 Likert scale
c_paper 1 - 5 Likert scale
c_tv 1 — 5 Likert scale

8d. “To judge what candidates are like, I use information from the following sources.”
d_cand 1 — 5 Likert scale
d_web 1 - 5 Likert scale

116

d_paper 1 - 5 Likert scale
d_tv 1 — 5 Likert scale

8e. “To remind me of my candidate’s strong points, I use information from the following
sources.”

a_cand 1 -— 5 Likert scale
a_web 1 - 5 Likert scale
e _paper 1 — 5 Likert scale
e_tv 1 — 5 Likert scale

8f. “To get information which agrees with my positions, I use information from the following
sources.”

f_cand 1 - 5 Likert scale
f_web 1 - 5 Likert scale
f _paper 1 — 5 Likert scale
f_tv 1 - 5 Likert scale

89. “To use what I learn in politics, I use information from the following sources.”
_cand 1 — 5 Likert scale

g_web 1 - 5 Likert scale
g_paper 1 — 5 Likert scale
tv 1 - 5 Likert scale

8h. “To give me something to talk about with other people, I use information from the following
sources.”

h_cand 1 - 5 Likert scale
h_web 1 - 5 Likert scale
h_paper 1 — 5 Likert scale

h_tv 1 - 5 Likert scale

8i. “Because it is interactive, I use information from the following sources.”
i_cand 1 - 5 Likert scale

i_web 1 - 5 Likert scale
i_paper 1 - 5 Likert scale
i_tv 1 - 5 Likert scale

8]. “Because it gives me the control over what and when I want to use, I use information from
the following sources.”

i_cand 1 - 5 Likert scale
j_web 1 - 5 Likert scale
j_paper 1 — 5 Likert scale
j_tv 1 - 5 Likert scale

8k. “To participate in the Presidential Election campaign, I use information from the following
sources.”

k_cand

k_cand 1 - 5 Likert scale
k_web 1 — 5 Likert scale
k _paper 1 - 5 Likert scale
k_tv 1 — 5 Likert scale
own

Do you have your own computer with lntemet access?
0 = No 1 = Yes

117

net_hour . Open-ended
How much time do you usually spend online per week?
Duration of the lntemet per week

computer 1 - 5 Likert scale

How comfortable are you in using a computer?
Computer comfortableness

lntemet 1 - 5 Likert scale

How comfortable are you in using the lntemet?
lntemet comfortableness

gender

0 = Female 1 = Male

age Open-ended

level

1= Freshman 2=Sophomore 3=Junior4=Senor 5=Graduate
major

1 = Humanities 2 = social science 3 = science 4 = other 5 = non-preference

Time 2 Survey Variables

vote2

Which candidate did you vote for on November 7?

1 = Bush 2 = Gore 3 = Nader 4 = Other 5 = Did not
Vote2x .

1 = Nader 0 = Bush 8 Gore 9 = Missing (undecided 8 other third party)
Voterx

1 = Third Party 0 = Bush 8 Gore 9 = Missing (undecided)

vote2y

1 = Any candidate 0 = Did not vote 9 = Missing

support2 1 - 5 Likert scale

Please mark the degree of support for the candidate you selected in the above question.

ifvote2

Only for those who “Did Not Vote,“ which candidate would you have voted for if you had had a
chance?

1 = Bush 2 = Gore 3 = Nader 4 = Other 5 = None

ifsup2 1 — 5 Likert scale
And mark the degree of support for the candidate you selected here.

118

Iogbush2 Open-ended ,

Ioggore2 Open-ended
Iognad2 Open-ended
Iogoth2 Open-ended

Al Gore (www.algore2000.com), George W Bush (www.georgewbush.com) and other
candidates have their official campaign web sites. Estimate how often you have visited to
each candidate official web site during the Presidential election campaign. If you have not
logged in to either candidate site, please write “0” in both.

L092

The aggregation of Iogbush2 through Iogoth2

Logb2bi

Ioggai

logn2bi

Base on Iongbush2, Ioggore2, and Iognad2, these three variable are recoded into binary
variables. 1 = logged on 0 = did not log

Logai .

Based on logz, whether or not one has logged in to any candidate site.
1 = at least 1 time 0 = Zero

logZeach Open-ended

Base on vote2bgn, IogZeach represents the number of logging on to the candidate web sites
whom respondents voted for.

newslogz Open-ended

In order to obtain information about the presidential election 2000, estimate the total number
of times you have spent to visit news web sites such as the New York Times (www.nyt.com)
and CNN (www.cnn.com).

anogai
Base on newslogZ, binary viriable
1= yes, logged on to news web sites 0 = no, did not log on to news web sites

conv1

Whether or not one converted from vote1 candidate to vote2 candidate.
1 = Bush -> Gore

2 = Bush -> Nader
3 = Bush -> Third

4 = Gore -> Bush

5 = Gore -> Nader
6 = Gore -> Third

7 = Nader -> Bush
8 = Nader -> Gore
9 = Nader -> Third
10= Third -> Bush
11 = Third -> Gore
12 = Third -> Nader

conv1bi
1 = converted . 0 = did not convert

conv2

119

Whether or not one converted from vote1 candidate to ifvote2 candidate.
1 = Bush -> Gore

2 = Bush -> Nader
3 = Bush -> Third

4 = Gore -> Bush

5 = Gore -> Nader
6 = Gore -> Third

7 = Nader -> Bush
8 = Nader -> Gore
9 = Nader -> Third
10= Third -> Bush
11 = Third -> Gore
12 = Third -> Nader

conv2bi
1 = converted O = did not convert

conv3

Whether or not one converted from'votel candidate to vote2 or ifvote2 candidate.
1 = Bush -> Gore

2 = Bush -> Nader
3 = Bush -> Third

4 = Gore -> Bush

5 = Gore -> Nader
6 = Gore -> Third

7 = Nader -> Bush
8 = Nader -> Gore
9 = Nader -> Third
10= Third -> Bush
11 = Third -> Gore
12 = Third -> Nader

conv3bi
1 = converted 0 = did not convert

sup 1_2

The difference between support1 and support2, not including the scores of coverts.
Positive sign = increased support

Negative sign = decreased support

0 = no change in support '

9 = missing (did not vote, converts)

sup 1_2x

1 = increased support

0 = decreased or unchanged support
9 = did not vote or convert

sup 1_2y

1 = increased support

0 = decreased support

9 = unchanged support, did not vote or convert

120

REFERENCES

121

REFERENCES

Abramovvitz, A.I. (1980). A comparison of voting for U.S. Senator and Representative.
American Political Science Review, 74, 633-640.

Agresti, A. & Finlay, B. (1997). Statistical methods for the social sciences (3rd edition).
Upper Saddle River, NJ: Prentice-Hall.

Ansolabehere, S.D., & Iyenger, S. (1995). Going negative: How political advertisements
shrink & polarize the electorate. New York: The Free Press.

Atkin, CK. (1972). Anticipated communication and mass media information-seeking.
Public Opinion Quarterly, 36, 188-199.

Atkin, C.K., Galloway, 1., & Nayman, QB. (1976). News media exposure, political
knowledge and campaign interest. Journalism Quarterly, 53, 231-237.

Atkin, C., Greenberg, B., Korzenny, F., & McDerrnott, S. (1979). Selective exposure to
television violence. Journal of Broadcasting, 23 (1), 5-13.

Bandura, A. (1986). Social foundations of thought and action: A social cognitive ap-
proach. Englewood Cliffs, NJ: Prentice Hall.

Becker, LB. (1979). Measurement of gratifications. Communication Research, 6 (1),
54-73.

Berelson, B.R., & Steiner, GA. (1964). Human behavior: An inventory of scientific find-
ings. New York: Harcourt, Brace & World.

Blumler, J .G. & McQuail, D. (1969). Television in politics: Its uses and influence. Chi-
cago: University of Chicago Press.

Brock, T.C., & Balloun, J .L. (1967). Behavioral receptivity to dissonant information.
Journal of Personality and Social Psychology, 6, 413-428.

Campbell, A., Converse, P.E., Miller, W.E., & Stokes, DE. (1960). The American voter.

122

New York: John Mley & Sons.

Cappella, J .N., & Jamieson, K.H. (1997). Spiral of cynicism: The press and the public
good. New York: Oxford University Press.

Cavanaugh, J .W. (1995). Media ejects on voters: A panel study of the 1992 presidential
election. Lanham, MD: University Press of America.

Center for Voter and Democracy. (2001). Voter turnout [Online]. Available:
http://www.fairvote.org/turnout/ July 24, 2001.

Chaffee, S.H. & Miyo, Y. (1983). Selective exposure and the reinforcement hypothesis:
An intergenerational panel study of the 1980 presidential election. Communica-

tion Research, 10 (1), 3-36.

Chaffee, S.H. & Kanihan, SF. (1997). Learning about politics from the mass media.
Political Communication, 14, 421 -430.

Cohen, A., Levy, M.R., & Golden, K. (1988). Chilch'en’s uses and gratifications of home
VCRs: Evolution or revolution. Communication Research, 15 (6), 772-780.

Cornfield, M. (1999a, May). The game of the name. Campaigns & Elections, 20(4), 3.

Cornfield, M. (1999b, June). Interacting for campaign dollar. Campaigns & Election, 20
(5), 31.

Cotton, J .L. (1980). Selective exposure to information and cognitive dissonance. Journal
of Research in Personality, 14, 518-527.

Cotton, J.L. (1985). Cognitive dissonance in selective exposure. In D. Zillm’an and J.
Bryant (Eds.), Selective exposure to communication (pp. 11-33), Hillsdale, NJ:

Lawrence Erlbaum.

Culbertson, H.M. & Stempel, G.H. (1986). How media use and reliance affect knowl-
edge level. Communication Research, 13 (4), 579-602.

123

Daft, R. & Lengel, R. (1984). Information richness: A new approach to managerial be-
havior and organization design. Research in Organizational Behavior' 6,
191-233.

Dafi, R. & Lengel, R. (1986). Organizational information requirements, media richness,
and structural design. Management Science 32, 554-571.

D’Alessio, D. (1997). Use of the World Wide Web in the 1996 US election. Election
Studies, 16 (4), 489-500.

DeFleur, M.L., & Ball-Rokeach, S. (1989). Theories of mass communication (5th edi-
tion). White Plains, NY: Longman.

Dennis, J., & Chafl‘ee, SH. (1978). .Legitimation in the 1976 U.S. election campaign.
Communication Research, 5, 371-394.

DiClemente, C.C., & Prochaska, J .0. (1985). Processes and stage of self-change: Coping
and competence in smoking behavior change. In S. Shifl‘man & T.A. Willis
(Eds.), Coping and substance abuse (pp. 319-343). San Diego, CA: Academic
Press.

Dickerson, J. (2000, January 31). Point, click, win!: Every candidate in the race has an
online strategy, but it’s still just the beta version of digital democracy. Time, 155,

42.

Ehrlich, D., Guttrnan, I., Schonbach, P., & Mills, J. (1957). Postdecision exposure to
relevant information. Journal of Abnormal and Social Psychology, 54, 98-102.

Feather, NT. (1962). Cigarette smoking and lung cancer: A study of cognitive disso-
nance. Australian Journal of Psychology, 1 4, 55-64.

Ferguson, D.A., & Perse, EM. (2000). The World Wrde Web as a functional alternative
to television. Journal of Broadcasting & Electronic Media, 44 (2), 155-174.

Festinger, L. (195 7). A theory of cognitive dissonance. Evanston, IL: Row Peterson.

124

Finkel. S.E., & Geer, J .G. (1998). A spot check: Casting doubt on the demobilizing effect
of attack advertising. American Journal of Political Science, 42 (2), 573-595.

Fisk, S.T. & Kinder, D.K. (1981). Involvement, expertise, and schema use: Evidence
from political communication. In N. Cantor and J .F. F ihlstrom (Eds.), Personal-
ity, cognition and social interaction (pp. 171-190). Hillsdale, NJ: Lawrence Erl-
baum.

Fisk, S.T., Lau, R.R., & Smith, R.A. (1990). On the varieties and utilities of political ex-
pertise. Social Cognition, 8(1), 31-48. I

Garramone, G.M., Atkin, C.K., Pinkleton, B.E., Cole, RT (1990). Effects of negative t,
political advertising on the political process. Journal of Broadcasting & Elec-
tronic Media, 34 (3), 299-311.

Graber, DA (1980). Mass media and American politics. Washington, DC: Congres-
sional Quarterly Press.

Grant, A.E., & Meadows, J .H. (2000). Communication technology update (7th editidn).
Wobum, MA: Butterworth-Heinemann.

Heider, F. (1946). Attitudes and cognitive organization. Journal of Psychology, 21,
107-112. '

 

Hill, K.A., & Hughes, J.E. (1998). Cyber politics: Citizen activism in the age of the
Internet. Lanham, MD: Rowman & Littlefield Publishers.

Hockaday, T. & Edlund, M. (1999, May). Banner advertising as a voter outreach tool.
Campaigns & Election, 20(4), 13-14.

Holtgrave, D.R., B.J. Tinsley, & L.S. Kay. (1995). Encouraging risk reduction: A deci-
sion-making approach to message design. In E. Maibach & R. L. Parrott (Eds.),
Designing health messages, (pp. 24-40). Thousand Oaks, CA: Sage

Hutcheson, G., & Sofroniou, N. (1999). The multivariate social scientist. London: Sage.

125